News & Updates
The latest news and updates from companies in the WLTH portfolio.
Mozilla fixes 271 Firefox vulnerabilities found by Anthropic's Claude Mythos in a single evaluation pass
Summary: Mozilla released Firefox 150 with fixes for 271 security vulnerabilities identified by Anthropic's Claude Mythos Preview, an unreleased frontier AI model distributed under the restricted Project Glasswing programme. The collaboration began with Claude Opus 4.6 finding 22 bugs in Firefox 148 earlier this year; Mythos produced more than twelve times as many. Firefox CTO Bobby Holley said the defects are "finite" and that defenders can "finally find them all," while the UK AI Security Institute confirmed Mythos can also execute autonomous multi-stage network attacks, making the dual-use tension the central policy question. Mozilla released Firefox 150 on Monday with fixes for 271 security vulnerabilities identified by Anthropic's Claude Mythos Preview, an unreleased frontier AI model restricted to a handful of organisations under Project Glasswing. The number is striking not because the bugs were exotic but because they were not. "We haven't seen any bugs that couldn't have been found by an elite human researcher," Mozilla said in a blog post titled "The zero-days are numbered." The point is that no human team could have found 271 of them this fast. The collaboration between Mozilla and Anthropic began earlier this year with a more modest effort. Starting in February, Firefox's security team used Claude Opus 4.6 to scan nearly 6,000 C++ files across the browser's codebase. That pass produced 112 unique reports, of which 22 were confirmed as security-sensitive bugs and shipped as fixes in Firefox 148. Fourteen were classified as high severity, representing almost a fifth of all high-severity Firefox vulnerabilities remediated in 2025. The Mythos evaluation, which followed as part of the continued partnership, produced more than twelve times as many confirmed vulnerabilities. Bobby Holley, Firefox's chief technology officer, described the experience as giving the team "vertigo." Claude Mythos Preview is the model at the centre of Anthropic's restricted Mythos model programme, Project Glasswing, announced on 7 April. It is a general-purpose frontier model, not a security-specific tool, but its coding capabilities have crossed a threshold that Anthropic considers significant enough to warrant controlled distribution. The UK's AI Security Institute evaluated the model and found it capable of executing multi-stage network attacks autonomously, completing a 32-step corporate network attack simulation called "The Last Ones" in three out of ten attempts. It can chain multiple small vulnerabilities into a single devastating attack, reconstruct source code from deployed software to find exploitable weaknesses, and build custom tools for lateral movement and data extraction once inside a network. Access is restricted to 12 named launch partners, including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, Nvidia, and Palo Alto Networks, with roughly 40 additional organisations granted access for defensive security work. Anthropic committed up to $100 million in usage credits and $4 million in direct donations to open-source security organisations, including $2.5 million to Alpha-Omega and OpenSSF through the Linux Foundation and $1.5 million to the Apache Software Foundation. The model is available to Glasswing participants at $25 per million input tokens and $125 per million output tokens through the Claude API, Amazon Bedrock, Google Cloud's Vertex AI, and Microsoft Foundry. The restricted rollout has already been tested. On the same day Anthropic announced Glasswing, a group of unauthorised users gained access to Mythos Preview by guessing the model's URL through a third-party vendor environment, an incident Anthropic said it is investigating. Holley framed the 271 vulnerabilities not as an indictment of Firefox's code quality but as evidence that the security landscape is shifting in favour of defenders for the first time. "A gap between machine-discoverable and human-discoverable bugs favors the attacker, who can concentrate many months of costly human effort to find a single bug," he wrote. "Closing this gap erodes the attacker's long-term advantage by making all discoveries cheap." The logic is straightforward. A zero-day vulnerability is valuable to an attacker precisely because it is unknown. If a defender can find and patch the same bug before an attacker discovers it, the bug has no offensive value. The cost asymmetry has historically favoured attackers: a browser like Firefox has millions of lines of code, and a single undiscovered flaw in any of them is enough for exploitation. An elite human security researcher might spend weeks or months finding one such flaw. A model like Mythos can scan the entire codebase in a fraction of that time. Mozilla's thesis is that this changes the economics permanently. "Software like Firefox is designed in a modular way for humans to be able to reason about its correctness," the blog post stated. "It is complex, but not arbitrarily complex. The defects are finite, and we are entering a world where we can finally find them all." The claim is bold and deliberately so. Mozilla is arguing that the age of zero-day vulnerabilities in well-structured software has an expiration date, not because attackers will stop looking, but because defenders will get there first. The 271 figure requires some unpacking. Mozilla's official security advisory for Firefox 150, MFSA 2026-30, lists 41 CVE entries, three of which are standard memory-safety roll-ups that aggregate multiple individual bugs under a single identifier. The 271 number represents the total count of discrete code defects identified by Mythos during its evaluation, many of which were grouped into those CVE bundles. The distinction matters because the headline number and the formal advisory number measure different things: one measures what the AI found, the other measures how much AI-generated code actually ships through the industry's standard vulnerability disclosure process. The most dangerous flaws include use-after-free vulnerabilities in the DOM and WebRTC components, the kinds of memory safety bugs that have been the bread and butter of browser exploitation for two decades. These are not novel attack surfaces. They are the same categories of bugs that Google's Project Zero has been finding across browsers since 2014. Google's own AI vulnerability research programme, Big Sleep, a collaboration between Project Zero and DeepMind, found a zero-day in SQLite in October 2024 and has since expanded to discover multiple flaws in widely used software. The difference with Mozilla's effort is scale: 271 bugs in a single evaluation pass, patched before release, across a codebase that has accumulated technical debt over more than two decades. The UK AI Security Institute's evaluation of Mythos Preview confirmed what the Mozilla results imply from the other direction: the same capabilities that make the model effective at finding vulnerabilities make it effective at exploiting them. The model became the first AI to complete "The Last Ones," a benchmark designed to simulate a full corporate network compromise. It succeeded in three out of ten attempts, averaging 22 of 32 steps across all runs. Independent testing confirmed that Mythos cannot reliably execute autonomous attacks against organisations with well-hardened defences, but the trajectory is clear. Each generation of frontier model has performed better on offensive security benchmarks than the last. This is the tension that Project Glasswing is designed to manage. By restricting Mythos to vetted organisations with defensive mandates, Anthropic is attempting to give defenders a structural head start, a window in which the good actors can scan and patch before the capabilities proliferate. The strategy depends on the restriction holding. The vendor breach on launch day suggests that containment is harder than access control. Anthropic has also identified thousands of zero-day vulnerabilities across every major operating system and every major web browser using Mythos, findings it is disclosing to the affected vendors through Glasswing. Anthropic's expanding enterprise footprint, from legal contract review in Microsoft Word to cybersecurity through Glasswing, reflects a company that is monetising Claude across every professional vertical where accuracy matters. The Mozilla partnership is the most dramatic demonstration yet, not because the model did something no human could do, but because it did what only a handful of humans can do, and did it 271 times in a single pass. Holley's conclusion captures both the promise and the vertigo: "Our work isn't finished, but we've turned the corner and can glimpse a future much better than just keeping up. Defenders finally have a chance to win, decisively." Whether that future arrives depends on whether the models that find the bugs remain in the hands of the people who fix them, or whether the capabilities leak faster than the patches ship. For now, Firefox 150 has 271 fewer ways to be broken. That is not a small thing. The question is how long that advantage lasts when the tool that found them is commanding extraordinary valuations precisely because of what it can do.
Anthropic Probes Alleged Unauthorized Access to AI Security Tool Mythos - IT Security News
The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user. The technical storage or access that is used exclusively for statistical purposes. The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
Credence Research Inc.: Liner Hanger System Market to Reach USD 6,268.42 Million by 2032, Driven by Deepwater Exploration and Unconventional Drilling Growth | Credence Research
LONDON, April 22, 2026 /PRNewswire/ -- Credence Research has released a new report titled "Liner Hanger System Market Size, Share, Growth, Opportunities, and Competitive Analysis, 2025-2032." According to the study, the Liner Hanger System Market was valued at USD 4,084.5 million in 2024 and is projected to reach USD 6,268.42 million by 2032, expanding at a CAGR of 5.6% during the forecast period. The market is witnessing steady expansion due to increasing drilling activity in unconventional reservoirs, growing investments in deepwater and ultra-deepwater projects, and rising demand for high-performance well completion technologies. The global liner hanger system market reflects a highly competitive and technology-driven environment, where major oilfield service providers focus on innovation, advanced materials, and automated deployment solutions. Companies such as Schlumberger Limited, Halliburton Company, Baker Hughes Company, Weatherford International plc, and NOV Inc. are strengthening their positions through high-pressure systems, expandable technologies, and digital monitoring capabilities. Regionally, North America led the market in 2024 with approximately 34% share, followed by Asia Pacific and Europe, supported by strong drilling programs across both onshore and offshore fields. Market Insights Growth is primarily driven by rising unconventional drilling activity and increasing deepwater exploration investments, which require advanced completion technologies.Conventional liner hangers held about 38% share in 2024, supported by their reliability and cost-efficiency across standard well operations.Key trends include growing adoption of expandable liner hangers and digital tools that enhance completion accuracy and reduce operational risks.North America dominates with around 34% share, followed by Asia Pacific (28%) and Europe (22%), supported by strong drilling and exploration programs. Market Overview: * Industry Landscape and Value Chain Assessment * Supply-Side Evaluation * Demand-Side Evaluation * Stakeholder Mapping * Porter's Five Forces Review * PESTLE Environment Assessment * Market Forecast and Future Direction * Short-Term Forecast (0-2 Years) * Mid-Term Forecast (3-5 Years) * Long-Term Forecast (5-10 Years) * Market Entry and Expansion Strategy Market Segmentation Analysis By Type Conventional liner hangers led the market in 2024 with about 38% share due to reliability and cost efficiency in standard wells. Mechanical and hydraulic types are gaining traction in deep and high-pressure wells, while expandable hangers grow in complex wellbores for better zonal isolation and larger internal diameter. By Well Type Horizontal and directional wells dominated with around 61% share in 2024, driven by unconventional drilling and longer laterals. Vertical wells maintain steady demand, but focus on higher recovery and efficiency continues to favor horizontal drilling. By Location of Deployment Onshore held nearly 63% share due to high drilling activity, lower costs, and easier operations. Offshore is growing steadily, especially in deepwater projects that require advanced high-pressure liner hanger systems. Request Free Sample Report (Liner Hanger System Market) - https://www.credenceresearch.com/report/liner-hanger-system-market Key Growth Drivers Rising Drilling Activities in Unconventional Reservoirs The expansion of shale gas, tight oil, and unconventional resource development is a major driver for liner hanger systems. These reservoirs require complex well geometries and multi-stage completions, increasing the demand for reliable hanger systems that maintain well integrity under high pressure. Horizontal drilling further strengthens this demand by requiring advanced solutions capable of stabilizing extended lateral sections and ensuring consistent performance. Expansion of Deepwater and Ultra-Deepwater Projects Growing investments in offshore exploration-particularly in regions such as the Gulf of Mexico, Brazil, and West Africa-are driving demand for high-performance liner hanger systems. These wells operate under extreme pressure and temperature conditions, requiring advanced technologies that ensure durability and long-term reliability. As offshore reserves become critical to global energy supply, demand for advanced completion systems continues to increase. Key Trends and Opportunities Shift Toward Expandable and High-Performance Systems Expandable liner hangers are gaining popularity due to their ability to maximize internal diameter and improve well efficiency. These systems support complex well architectures and reduce clearance issues, making them ideal for advanced drilling environments. Manufacturers are investing in stronger alloys and corrosion-resistant materials to meet growing demand for high-performance solutions. Digitalization and Automation in Well Completion Automation and digital technologies are transforming liner hanger deployment processes. Real-time monitoring, predictive analytics, and automated running tools improve accuracy, reduce human error, and enhance operational efficiency. These innovations create opportunities for technology-driven product development and improve overall well completion performance. Key Challenges High Operational Costs in Complex Wells Deepwater and unconventional wells require advanced liner hanger systems that involve high procurement and installation costs. These costs can limit adoption, particularly for smaller operators or during periods of oil price volatility. Technical Failures in Extreme Conditions Liner hanger systems operate under high stress, temperature, and corrosion conditions, increasing the risk of mechanical failures. Ensuring consistent reliability remains a challenge, requiring continuous investment in material innovation and testing standards. Regional Analysis North America North America held the largest market share in 2024 at around 34%, driven by strong shale drilling activity in the U.S. and Canada. High investment in horizontal drilling, multi-stage completions, and mature field redevelopment supports continuous demand for advanced liner hanger systems. Europe Europe accounted for approximately 22% share, supported by offshore activities in the North Sea and strict well safety regulations. The region emphasizes high-reliability systems to operate in harsh offshore conditions. Asia Pacific Asia Pacific captured nearly 28% share, driven by expanding exploration programs in China, India, and Southeast Asia. Rising energy demand and increased offshore drilling investments continue to support regional growth. Latin America Latin America held about 10% share, with strong demand from Brazil's deepwater fields and Argentina's unconventional resources, particularly Vaca Muerta. Middle East & Africa The Middle East & Africa accounted for around 6% share, supported by extensive drilling activities and rising offshore developments across Saudi Arabia, UAE, and West Africa. Competitive Landscape JC PetroSchlumberger LimitedInnovex Downhole Solutions, Inc.Weatherford International plcNOV Inc.NCS MultistageDril-Quip, Inc.Allamon Tool CompanyPackers Plus Energy Services Inc.Baker Hughes CompanyHalliburton Company Request Free Sample Report (Liner Hanger System Market)- https://www.credenceresearch.com/report/liner-hanger-system-market Related Reports - Coiled Tubing Services Market https://www.credenceresearch.com/report/coiled-tubing-service-market Coal Bed Methane Market https://www.credenceresearch.com/report/coal-bed-methane-market Angola Oil And Gas Upstream Market https://www.credenceresearch.com/report/angola-oil-and-gas-upstream-market Pipeline Integrity Management Market https://www.credenceresearch.com/report/pipeline-integrity-management-market Pipeline Monitoring System Market https://www.credenceresearch.com/report/pipeline-monitoring-solutions-market Spearmint Oil Market https://www.credenceresearch.com/report/spearmint-oil-market Sustainable Aviation Fuel Market https://www.credenceresearch.com/report/sustainable-aviation-fuel-market Marine Growth Removal (MGR) Market https://www.credenceresearch.com/report/marine-growth-removal-mgr-market Alkylate Market https://www.credenceresearch.com/report/alkylate-market Oil And Gas Actuators Market https://www.credenceresearch.com/report/oil-and-gas-actuators-market About Us Credence Research Inc is a global market intelligence and consulting firm founded in 2015. It delivers deep market insights, quantitative analysis, and strategic guidance to business leaders, investors, governments, NGOs, and non-profit groups worldwide. The company helps organizations evaluate markets, understand trends, reduce risk, and make data-driven decisions that support growth and competitive strategy. Credence Research is known for rigorous research methods and comprehensive analytics. The firm produces detailed reports covering market size, forecasts, growth drivers, trends, and competitive landscapes across many industries. Each report often includes frameworks like PESTLE and Porter's Five Forces to give a complete view of market dynamics and future potential. Credence Research also provides tailored consulting services, due diligence support, go-to-market planning, and pre-IPO research to strengthen client strategies and investment narratives. Its insights come from both primary and secondary research, expert interviews, and advanced data modelling. The firm's client base spans Europe, the Americas, Asia-Pacific, and the Middle East/Africa. To find out more, visit www.credenceresearch.com or follow us on X.com, LinkedIn and Facebook Contact Us - Credence Research Inc, Tower C-1105, S 25, Akash Tower, Vishal Nagar, Pimple Nilakh, Haveli, Pune - 411027, India USA: +1-888-600-6441 Email: [email protected] Visit Our Website: https://www.credenceresearch.com/ Logo - https://mma.prnewswire.com/media/2562161/Credence_Research_Logo.jpg View original content:https://www.prnewswire.co.uk/news-releases/liner-hanger-system-market-to-reach-usd-6-268-42-million-by-2032--driven-by-deepwater-exploration-and-unconventional-drilling-growth--credence-research-302750716.html © 2026 PR Newswire
SpaceX Says It Can Buy AI Coding Tool Cursor for $60B Later This Year
SAN FRANCISCO (AP) -- SpaceX says it has the rights to buy artificial intelligence coding tool Cursor for $60 billion later this year as Elon Musk's space exploration and AI company looks for ways to compete with rivals Anthropic and OpenAI ahead of a planned Wall Street debut. SpaceX said that, alternatively, it could pay $10 billion to "work together" with Cursor. SpaceX announced the deal Tuesday on the social platform X, which along with the AI chatbot Grok is part of a constellation of properties that Musk has merged into his rocket company. Cursor, made by San Francisco startup Anysphere, is a popular AI coding assistant. What SpaceX describes as Cursor's wide "distribution to expert software engineers" is likely part of what makes it attractive to Musk's company, giving it access to a new customer base. Cursor said its new partnership with SpaceX subsidiary xAI will enable it to build future AI products using xAI's massive AI data center complex Colossus, based in Memphis, Tennessee. "We've wanted to push our training efforts much further, but we've been bottlenecked by compute," Cursor said in a statement on X, which didn't mention the possibility of being acquired. "With this partnership, our team will leverage xAI's Colossus infrastructure to dramatically scale up the intelligence of our models." Cursor, which started in 2022, helped sparked a trend called "vibe coding" as AI coding assistants have become increasingly capable of doing the work of computer programming. Cursor competes with other coding tools like Anthropic's Claude Code and OpenAI's Codex but also has relied heavily on partnerships with those larger AI research companies for the foundations of its technology. It was Cursor's Composer, combined with Anthropic's Claude Sonnet, that a prominent AI researcher was playing with for weekend projects when he coined the phrase "vibe coding" in early 2025.
Anthropic's Locked-Down Mythos Model Hit by Access Claim | This Week in IT - Techopedia
Suswati Basu is a multilingual, award-winning editor. She was shortlisted for the Guardian Mary Stott Prize and longlisted for the Guardian International Development Journalism Award.... Anthropic is investigating reports that Claude Mythos Preview, an unreleased version of its AI model, may have been accessed without authorization through a third-party vendor environment tied to development work. Speaking to Techopedia, an Anthropic spokesperson said: "We're investigating a report claiming unauthorized access to Claude Mythos Preview through one of our third-party vendor environments." Only a week after releasing the Claude Mythos Preview to a select group of organizations, people familiar with the matter said the reported activity appears linked to an external development platform rather than Anthropic's production API systems. The sources added there is no evidence at this stage that the incident extended beyond that external environment or affected the company's internal infrastructure. Anthropic's Mythos Model Is Already Being Put To The Test Anthropic has not said whether any data was removed or when the alleged access took place, however, Bloomberg reported that it took place on a "private online forum." The users were reportedly part of a private Discord group focused on uncovering details about unreleased AI models, using bots to scan unsecured websites such as GitHub for stray references posted by major labs. The news outlet reported that the group gained access to Mythos after some members made an educated guess about the model's online location using naming patterns Anthropic had used for earlier releases. Some of those clues allegedly emerged from a recent data breach involving Mercor, a startup that works with several leading AI developers. The irony is hard to miss. Anthropic positioned Mythos as a model so powerful that it required an unusually cautious rollout. It limited access to a small number of trusted partners because of fears it could be misused by hackers or destabilize cybersecurity defenses. Yet the first major controversy surrounding the system is not what Mythos itself might do. It's the possibility that third parties have already gained access through the carelessness of an external development partner. For a company that has built much of its identity around AI safety and controlled deployment, this risks reinforcing a familiar lesson in tech: A system is only as strong as the weakest link in its wider supply chain. Quite often, that weak link is basic human nature. Also in Tech News Tim Cook Steps Down As Apple Addresses Its AI Problem After more than a decade at the helm, Apple's head honcho Tim Cook passed the baton to John Ternus, signaling a change in direction for the $4 trillion company. In a statement, the 65-year-old said Ternus would attempt to "make something better, bolder, more beautiful, and more meaningful." Ternus has been serving as the tech giant's senior vice president of Hardware Engineering. The changing of the guard comes at a time when Apple appears to have stalled in the AI race against the likes of OpenAI, Google, and Grok. Cook's tenure as CEO will end on September 1, bringing to an end an era defined by operational efficiency and financial growth Although he ushered Apple into its trillion-dollar era, Cook has often lived in the shadow of his predecessor. Analysts have built a mythos around company cofounder Steve Jobs, next to whom Cook has seemed perhaps too straight-laced. Now, Ternus will be expected to step up as both a master of managing sprawling operations and an innovation wizard for this new tech era. Ming-Chi Kuo, a tech analyst at TF International, wrote on X that one of Ternus's major achievements was overseeing the transition from Intel processors to the firm's own proprietary silicon. Kuo added: "Without this, there wouldn't be the success of today's MacBook Neo or the advantage Apple now holds as it gears up for AI devices." Meta Plans to Track Employee Keystrokes for AI Training Meta has found itself in hot water after reports emerged that it plans to track the computer activity of U.S. employees to help train its AI models. The software is expected to capture mouse movements, clicks, and keystrokes as the company looks to build AI agents capable of working more autonomously, Reuters first reported, citing an internal memo. According to the report, the company's Model Capability Initiative tool would run across work-related apps and websites, while also taking occasional snapshots of content displayed on employees' screens. Techopedia contacted Meta for comment, but an initial email bounced back. We will continue to seek a response. The move has already drawn criticism from privacy and ethics experts. Veith Weilnhammer, a Max Planck Fellow in Computational Psychiatry, wrote on LinkedIn: "Beyond questions about AI systems that emulate human behavior, such as their impact on the job market, privacy, and the growing commercial value of human behavioral knowledge, this raises an important societal issue: How should we govern access to human-computer interactions, and how can these data be used for public good?" For now, the data collection is reportedly limited to the U.S., with stricter privacy rules likely to make a similar rollout more difficult in Europe. UK Cyber Chief Warns Frontier AI Is Accelerating Exploit Discovery Britain's top cybersecurity official is expected to warn that frontier AI models are making it easier to discover and exploit software flaws at scale, as the UK confronts a rising mix of technological disruption and geopolitical threats. In remarks due to be delivered at the CYBERUK conference in Glasgow on Wednesday (April 23), National Cyber Security Centre chief executive Richard Horne is set to say that while AI has the potential to strengthen cyber defense, adversaries will also move quickly to weaponize the technology. Politico reported Horne will caution that frontier AI is already "rapidly enabling discovery and exploitation of existing vulnerabilities at scale," increasing pressure on organizations to patch systems, replace legacy technology, and improve basic cyber hygiene. Researchers said Anthropic's Mythos, for example, was too dangerous for general release because of its alleged ability to help users identify and exploit sophisticated vulnerabilities. And just like that, we've gone full circle back to Anthropic.
There's No Time for SpaceX to Buy Cursor
Traditionally the way an initial public offering works is that there is a private company, and it wants to sell its stock to public investors, so it goes out and markets itself to those investors. Along with its bankers and lawyers, the company writes a prospectus explaining its business. The prospectus will contain some standard sections: often a manifesto from the company's founder setting out its vision and guiding philosophy, an overview of its business and the competitive landscape, a discussion of its financial results. There will be audited historical financial statements going back several years, and risk factors explaining what might go wrong. There will also normally be a roadshow, in which the company and its advisers present the company to investors in meetings or over Zoom. The roadshow presentation will hit some of the same notes as the prospectus, though it will probably be more forward-looking, more about how much money the company will make next year than how much it made last year, more about plans than about risks. Research analysts at the company's banks will explain their -- preferably optimistic -- views on the company to potential investors. Salespeople at the banks will call their customers to sell them stock. SpaceX, Elon Musk's satellite internet, rocket launch, space data center, Mars colonization, frontier AI model and social media company, plans to do an IPO sometime soon, possibly in June. On the one hand, this seems like a complicated job: As I hinted in the previous sentence, it is not easy to describe what SpaceX does, what its guiding philosophy is, how it makes money, what risks it faces, or what businesses it competes in. What are the EBITDA margins on the space data centers? What sorts of contingent liabilities does the Mars colony face? The prospectus -- essentially conservative and backward-looking -- will not capture the true meaning of SpaceX; presumably it will barely mention the Mars colonies or space data centers, which are at this point entirely hypothetical. The roadshow will mention those, but in a fairly abstract way. It's not like there are meaningful projections of 2029 space data center revenue. On the other hand, this is stupid, I'm kidding, and none of this stuff matters for SpaceX. The SpaceX IPO is not really about a private company introducing and explaining itself to the general public. The SpaceX IPO is about Elon Musk Elon Musking the Elon Muskiest of Elon Musk. "SpaceX's IPO Pitch Centers on Elon Musk's Ability to 'Sell the Dream'" was a Bloomberg headline a few weeks ago: "What you've got to be convinced of -- and this is what they'll be working on until this is filed publicly -- is continuing to sell the dream and basically there's nobody that's been better at selling the dream than Elon Musk," said David Erickson, an adjunct associate professor at Columbia Business School and a former co-head of global equity capital markets at Barclays Plc. ... "The reality is it's not about the fundamentals, nobody is going to get there on the fundamentals from a math standpoint because the math doesn't work," said Erickson. It is a vibes-based IPO. And, you know, fine. There is a long history of investors giving Elon Musk a lot of money on faith, assuming that he'll do good stuff and it will all work out for them, and it usually does. When Musk bought Twitter Inc. in 2022, various rich people rather cavalierly chucked in some of their own money to invest in the deal. This worked out incredibly well, for them, and Twitter is now part of SpaceX. Part of what you are getting, when you invest in Elon Musk's Whole Thing, is exposure to his restlessness. SpaceX was a company founded on the principle that it would be cool to shoot rockets into space. Musk still believes that, but in the years since SpaceX's founding he has added many more visions -- brain implants, tunnels, artificial superintelligence, data centers, exposing millions of people to his online comedy stylings and political opinions -- to his portfolio. (Also he runs a car company.) SpaceX combines many (not all!) of those visions, but today's portfolio of SpaceX businesses is to some extent accidental. Today I can describe SpaceX as a "satellite internet, rocket launch, space data center, Mars colonization, frontier AI model and social media company," but it will be somewhat shocking if I can use the same list in April 2027. In 2027 most of those things will still be on the list, but something else, something that would never have occurred to me, will be occupying much of Musk's and SpaceX's time and attention. "Humanity's survival depends on ______," Musk will say, about this surprising new thing, "and SpaceX is pivoting all of the computing power in its space data centers to solve it." Space data centers! Musk's restlessness fits right in to a vibes-based, dream-selling IPO: He can show up at the roadshow in June and be like "by this time next year the space data centers will be made out of a previously undiscovered element that we will acquire from aliens," and investors will be like "oh man is he ever selling the dream." If you're investing in Musk you want novelty; if the roadshow was just like "satellite internet is a good steady business" you would be disappointed and would not pay 100 times revenue for the stock. But it fits terribly into the traditional IPO process, the kind with a prospectus and risk factors and historical financial statements. You can't pivot your business every two months while also running an IPO process; there is a long lead time to get audited financial statements, lock down the prospectus, have it reviewed by the US Securities and Exchange Commission, educate the bankers and salespeople, and sell the stock. In some sense the prospectus for SpaceX should just be a headshot of Elon Musk with "trust me!" scrawled on it with a glitter pen, but the SEC wouldn't approve that. Anyway. One important and quite recent part of the SpaceX dream is xAI, Musk's frontier AI lab. The nice thing about Musk's restlessness is that he was able to build a highly competitive AI lab starting from scratch, quickly competing with established AI pioneers like Google and OpenAI. The bad thing about his restlessness is that then he tore it up: Most of his xAI co-founders have left and he is restarting from scratch. "xAI was not built right first time around, so is being rebuilt from the foundations up," Musk tweeted, IN MARCH, like THREE MONTHS before SpaceX is supposed to go public. One way to speed that process up is, apparently, to buy Cursor, an AI coding company. "SpaceX Has Deal for Right to Acquire Cursor for $60 Billion," reports Bloomberg News, and we'll come back to that phrasing in a minute. "The combination of Cursor's leading product and distribution to expert software engineers with SpaceX's million H100 equivalent Colossus training supercomputer will allow us to build the world's most useful models," SpaceX tweeted today. At Stratechery, Ben Thompson explains the rationale: SpaceXAI has a ton of compute, and no one to use it, either for R&D or inference. Cursor, meanwhile, reached a peak of its own a year ago, when it was the only AI application other than ChatGPT earning meaningful revenue. It's a real product with real customers, in a real trap: Cursor has to pay the frontier labs for API access to their models, which themselves offer heavily subsidized and increasingly agentic coding products on a subscription basis. The end result is a startup with real traction that is both bleeding money and leading edge customers. ... There is really obvious synergy between SpaceXAI and Cursor: the former has compute, and the latter has a product, data, and a decent amount of distribution for the use case that is most important for AI. At a high level this tie-up makes a lot of sense. Fine, okay, right, last week SpaceX was betting its future on xAI's AI capabilities, and this week it's betting its future on Cursor's AI capabilities. That's how life works in Elon Musk's Whole Thing, and his investors wouldn't have it any other way. But the IPO! There's an IPO! In like two months! It's bad enough that the SpaceX IPO became Also The xAI And Twitter IPO in February, but making it also the Cursor IPO now is too much. There is no time to acquire Cursor before the IPO. So it will have to wait. "Cursor has also given SpaceX the right to acquire Cursor later this year for $60 billion or pay $10 billion for our work together," SpaceX's tweet says. Bloomberg reports: SpaceX isn't acquiring Cursor immediately because of the rocket company's imminent initial public offering, according to a person familiar with the matter, who asked not to be identified discussing nonpublic information. A major transaction would require SpaceX to update its filings and financial details, potentially delaying the IPO, which is targeting a $2 trillion valuation. The $10 billion is a breakup fee if the deal doesn't go through, according to people with knowledge of the deal. I would put it a bit differently: The $10 billion is option premium, giving SpaceX the right, but not the obligation, to acquire Cursor "later this year," after the IPO stuff has calmed down.1 Because SpaceX doesn't own Cursor, or even have a binding agreement to acquire it, the IPO prospectus probably does not have to get into too much detail on Cursor's business or finances or how they would combine with SpaceX's. The SpaceX IPO prospectus will be a historical document, capturing Elon Musk's Whole Thing at a specific moment in time, but everyone understands that, moments later, things will change. Cursor is part of that "moments later" bucket: Everyone has fair warning that, after the IPO, SpaceX will also be Cursor. But nobody has much detail about what that will entail, in the same way that nobody has much detail now about the advanced technology that Musk will license from aliens when he finds them, or about whatever he gets up to next.2 Sure SpaceX could wait to buy Cursor, but: * AI is a competitive and fast-moving market and it can't wait three months to start the integration.3 Formally, the acquisition will happen after the IPO, but "SpaceXAI and @cursor_ai are now working closely together to create the world's best coding and knowledge work AI." * If you're leaving your startup to go work for Musk, a famously demanding and mercurial boss, you will want to get cashed out of your startup. Selling for $60 billion is a good deal; going to work for him on spec for a few months is not. * Given Musk's mergers-and-acquisitions track record, a pinky swear that he will eventually buy the company is not sufficient. You will want, say, $10 billion of committed option premium to make it worth your while. Cursor's management learned from what Twitter went through. "Among the largest termination fees in history," the Financial Times called it, which is what you should expect in a post-Twitter Musk acquisition. * Also, I mean, you can't incorporate Cursor's business in the IPO prospectus, but you can mention it. "We will own Cursor and build a good coding product" is fair game for the roadshow; it's part of selling the dream. Adding specific financial disclosure to the IPO, at this point, would be bad; adding cool new things to talk about is good. So this approach makes sense. The deal is certain enough to satisfy Cursor and Musk and SpaceX's potential investors, but uncertain enough to satisfy SpaceX's accountants and lawyers and the SEC. Of course Musk does change his mind a lot. It would be very funny if he sours on Cursor by July and walks away from the deal, and they make $10 billion for three months' work. I guess that's another reason not to spend much time on Cursor in the prospectus. People are worried about private market liquidity One point that I have made, during the recent private credit troubles, is that defined-benefit pension funds are the ideal investors in private markets. People mostly invest for retirement, retirement is a long way away for many of them, and so they should, statistically, be willing to earn a premium by investing in long-term illiquid assets. Pension funds, historically, did that: Everyone pooled their money for retirement, and the fund invested it in long-term stuff. The fund's obligations -- pay $X of benefits in Y years -- were predictable and long-term enough that it could lock up its money. The modern American approach of mostly defined-contribution 401(k) retirement savings blows this up, and the main reason private credit funds are having problems is that they are sold directly to retirement savers rather than to pension funds. That is my perspective as an American: I have a 401(k), and defined-benefit pensions are pretty uncommon around here. But I should not idealize them too much. We talked back in 2022 about a quirk of the UK pension experience, in which, for an assortment of accounting and economic reasons that I explained at the time and won't repeat here, UK defined-benefit pension plans have become enormous users of short-term leverage to buy gilts (UK government bonds). This is called "liability-driven investing," or LDI. Basically, if interest rates go up, the pension plans get margin calls and have to sell assets to post more collateral to support their leveraged gilt positions. That seems ... bad? I wrote: The thing that was so good about pension funds -- their structural long-termism, the fact that you can't have a run on a pension fund: You've ruined that! Now, if interest rates go up (gilts go down), your bank will call you up and say "you used our money to buy assets, and the assets went down, so you need to give us some money back." And then you have to sell a bunch of your assets -- the gilts and stocks that you own -- to pay off those margin calls. Through the magic of derivatives you have transformed your safe boring long-term pension fund into a risky leveraged vehicle that could get blown up by market moves. I know this is bad but I find something aesthetically beautiful about it. If you have a pot of money that is immune to bank runs, over time, modern finance will find a way to make it vulnerable to bank runs. That is an emergent property of modern finance. No one sits down and says "let's make pension funds vulnerable to bank runs!" Finance, as an abstract entity, just sort of does that on its own. One way to think of it is that pensions were invented, and they were a good stable long-term way to save for retirement, and over time finance inexorably made them run-prone, but in different ways. In the UK, pensions became run-prone by using LDI; in the US, they became run-prone by becoming 401(k)s. Anyway, in the US, people are worried about private market liquidity because private-market funds have increasingly been stuffed into retail investors' portfolios, and retail investors get nervous and try to sell. In the UK, people are worried about private market liquidity because private-market funds tend to be owned by pension funds, and the pensions might get nervous. The Financial Times reports: UK pension funds have been warned they face "huge" costs if they try to offload their private market assets, following a warning by the industry regulator about some schemes' high exposure to hard-to-sell investments. In response to a letter sent late last year by The Pensions Regulator requesting that schemes review their liquidity positions, some trustees have been advised by their consultants not to sell illiquid assets because of the "huge penalties" involved, according to people familiar with the schemes. ... TPR wrote to the trustees of 58 funds in November, warning that more than 50 per cent of each of their portfolios were invested in assets that would take longer than a week to sell. "If you have not done so already, we would encourage you to review [your] scheme's liquidity position . . . to assess the robustness of [your] investment strategy," it wrote. Having easy-to-sell assets available "is important in maintaining the investment strategy if interest rates rise materially and there are collateral calls . . . [and] in making benefits payments". Regulators have sharpened their oversight of defined benefit pension schemes since their forced selling of UK government bonds by so-called liability-driven investment strategies triggered chaos in the gilt market after Liz Truss's "mini" Budget of September 2022. The crisis, and the TPR's subsequent requirement for pension schemes to hold more in easy-to-sell assets, left many overexposed to illiquid assets, with some still struggling to rebalance. In a perfect world, defined-benefit pension funds would have predictable long-term liabilities, which they could match by investing in illiquid long-term private assets. But we don't live in that world. Bespoke parametric insurance If you live in a flood zone, you might want to buy insurance against the risk that there will be a big rainstorm and your house will be washed away. Here are two ways to think about that insurance: * Someone sells you a policy that is like "if a flood damages your house, we'll pay your cost to repair it." * Someone sells you a policy that is like "if your town gets more than X inches of rain in a 24-hour period, we'll pay you $Y." The first policy -- one that pays you based on your actual loss -- is simpler for you. It exactly covers your risk. The thing you care about is your house being damaged, and you get paid if your house is damaged. The second policy pays you for an input, for a factor in your house being damaged. But sometimes the town will get fewer than X inches of rain and your house will be damaged, in which case the insurance won't pay out but you'll wish it would. Other times, the town will get more than X inches of rain and your house will be fine, in which case the insurance payout is just a windfall. You have basis risk; the insurance payout doesn't perfectly correspond to the risk. The second policy, though -- the one that pays you based on rainfall -- is simpler for the insurance company. It's easier to underwrite: The insurance company can have some general rainfall model for your town, and sell everyone in town insurance at the same rates. It doesn't have to individually evaluate each house's location and construction. It is also easy to administer claims: If it rains more than X inches, everyone gets paid; if not, no one does. The company doesn't need to go to each house and check for damage, or argue over whether you were at fault for not closing your windows. Just a simple statistical payout. This also has some advantages for you. For one thing, it might be cheaper. But also, if there is a storm, the insurance company doesn't have to send someone out to assess your damages and haggle over your claim; if the rainfall trigger is hit, they just write you a check. And so in fact most normal insurance is loss-based -- you get paid based on your actual losses -- but there are some important cases of "parametric insurance," where you get paid for some amount of rainfall. Catastrophe bonds, for instance, are sort of a distributed market-based form of insurance, often issued by countries where it is hard to assess claims, so they will often have parametric triggers that pay based on air pressure or rainfall rather than damages. Neither approach is perfect, though. What you might want is to combine the simplicity and objectivity of parametric insurance with the specificity of loss-based insurance. "If there's more than X inches of rain directly on my roof, pay me $Y." With more sensors everywhere, and with more sensitive weather models, that seems doable. Bloomberg's Joe Wertz and Mary Hui reported on Monday that Bad Bunny did it: Bad Bunny had a problem. Days before the Puerto Rican mega-star was set to perform in Colombia earlier this year, the possibility of heavy rains threatened to turn three sold-out shows in Medellin into multimillion-dollar losses. Solutions were few. Traditional event cancellation insurance is rarely available so close to a performance date. Even if a policy could be secured, geography posed another challenge. The nearest official weather sensor sat a mile from the open-air venue -- too far, in a tropical city with steep terrain and complex microclimates, to serve as a reliable trigger for a payout. So a transatlantic team of brokers, underwriters and meteorologists devised a workaround to save Bad Bunny from losses: They installed a temporary weather station inside the stadium, linked to a bespoke policy that would pay out if rainfall exceeded a set threshold. And: For [Vaisala Oyj, the weather monitoring company] and Descartes [Underwriting SAS, the insurance company], the temporary weather station was a novel fix for what insurers call "basis risk" -- the chance that rain falls at the venue but isn't recorded at an official weather station used to settle the contract. Putting the station inside the stadium solved that problem. ... In that context, the Colombia concert was a successful test run for both companies, which see more untapped demand for bespoke parametric insurance paired with on-site weather monitoring. Sure, right, eventually we will all have our own weather stations and our own bespoke parametric insurance. Nothing is securities fraud I mentioned yesterday that SpaceX, when it becomes a public company in the next few months, apparently plans to prohibit shareholder lawsuits: If you buy SpaceX stock, and you have any complaints, you have to take it up with "mandatory arbitration," not a lawsuit (and, thus, not a class-action lawsuit). US public companies have not historically been allowed to force shareholders into arbitration, and so shareholders sue often, enthusiastically and in class actions. "Everything is securities fraud," I like to say. But last year the SEC changed its views on this, so now companies can require arbitration. "Nothing Is Securities Fraud?," I said. But yesterday I wrote: "As far as I can tell, no one has done it yet. Now SpaceX might." That was wrong. The first company to do it seems to have been Zion Oil & Gas, in December. Good for them, and sorry for missing them. That said, I also wrote yesterday that "if SpaceX does it and it works, then 100 companies that are not run by Elon Musk are going to try." And that's probably right. As, uh, the state of Texas could tell you, Elon Musk's corporate structure decisions tend to be closely watched and widely followed. Speaking of which: I also mentioned yesterday that SpaceX plans to go public with two classes of shares, with Musk getting super-voting shares so he can control the company forever while owning only a minority of its economic value. I made three points: * Yes, right, that's obviously how a Musk company should work, * Tesla Inc. should have done that when it went public, and * Tesla can't do it now, because stock exchange rules don't allow already-public companies to start issuing super-voting shares. So to increase Musk's voting control, Tesla has to -- and will! -- give him a much bigger economic slice of the company. Several readers emailed to suggest the obvious solution here. If SpaceX (with its dual-class stock) acquires Tesla, then Musk can probably get voting control of both companies forever and also, you know, have a fun time doing a merger. So look out for that in like six months. Right after the Cursor deal closes. Things happen More Banks Are Earning $100 Million Fees for Advising Big M&A Targets. Job Cuts Driven by A.I. Are Rising on Wall Street. OpenAI in talks to commit up to $1.5bn to private-equity joint venture. OpenAI Is Working With Consultants to Sell Codex. As Blue Owl's Strains Worsened, Its Bankers Pitched a Long Shot. Private Credit BDCs' 2028 Maturity Wall Poses Risk, Moody's Says. Trump Administration Nears $500 Million Spirit Rescue Package. Anthropic's Mythos Model Is Being Accessed by Unauthorized Users. Insurers move to cap cyber payouts related to AI and 'LLMjacking.' Unraveling Hedge Fund Trade Derails Boom in Taiwan Convertibles. DOJ Charges Southern Poverty Law Center With Bank Fraud. AI Startup Has Helped Reverse Thousands of Denied Health Insurance Claims. Elite law firm Sullivan & Cromwell admits to AI 'hallucinations.' Nine West is back. Lenox Millionaires. War disruption forces world's biggest condom maker to raise prices by up to 30%. If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!
Latest shot on Anthropic's Mythos -- China's cybersecurity giant Qihoo 360 finds 1,000 software vulnerabilities fast, raising global zero-day risks
China's Qihoo 360 has flagged nearly 1,000 software vulnerabilities using AI in record time. This is a major cybersecurity shift. AI vulnerability scanning is now faster than ever. It is changing global cyber defense rules. The findings come soon after advances linked to Anthropic's Mythos systems. The 360 Digital Security Group is using an AI-driven vulnerability discovery agent to scan software like Microsoft Office and modern AI frameworks. Experts warn of rising zero-day vulnerabilities and real-time exploit risks. Hackers can act faster now. Security teams must respond instantly.
SpaceX Admits AI Data Centers in Space May Be a Really Terrible Idea
SpaceX CEO Elon Musk has been adamant that putting enormous AI data centers in Earth's orbit is a "no-brainer." At the World Economic Forum in Davos earlier this year, Musk claimed that the unconstrained solar power beyond our planet's surface makes it the "lowest-cost place to put AI," something "that will be true within two years, three at the latest." Yet many experts remain unconvinced that sending up to a million satellites, each bigger than the International Space Station, makes any sense, citing concerns over economic feasibility and physical limits. Some warn they could even cause an environmental catastrophe, with aging and failing hardware doomed to burn up in the Earth's atmosphere while releasing copious amounts of ozone-depleting chemicals. Even SpaceX itself isn't fully sold on the idea its mercurial CEO is pushing hard -- and prioritizing over his decades-old wish of sending humans to Mars. In excerpts of the company's pre-IPO filing obtained by Reuters, the company admitted that its "initiatives to develop orbital AI compute and in-orbit, lunar, and interplanetary industrialization are in early stages, involve significant technical complexity and unproven technologies, and may not achieve commercial viability." It's a significant tonal shift ahead of the company's blockbuster IPO, which is reportedly on track for later this year, at a record-breaking valuation of roughly $1.75 trillion. SpaceX acknowledged that sending sensitive AI chips into space may cause them to wear out much faster. Orbital data centers will operate "in the harsh and unpredictable environment of space, exposing them to a wide and unique range of space-related risks that could cause them to malfunction or fail." The company also pointed out its heavy reliance on its super heavy launch platform, Starship, to build out its enormous constellation of up to one million data centers. "Any failure or delay in the development of Starship at scale or in achieving the required launch cadence, reusability and capabilities thereof would delay or limit our ability to execute our growth strategy," the filing reads, as quoted by Reuters. SpaceX has yet to successfully send the rocket into space and safely land it in one piece. Previous test flights have resulted in massive mid-flight explosions. However, thanks to its enormous size and payload capacity, SpaceX could soon dramatically increase the amount of hardware it can launch into space, while theoretically lowering costs as well. In short, its plans for orbital data centers aren't just extremely ambitious; Musk has framed his entire space company on the endeavor months ahead of its long-awaited IPO. Yet glaring questions regarding the idea's financial, let alone physical, feasibility remain unanswered -- a reality that the company's engineers are clearly having trouble ignoring.
Meet Aman Sanger: Indian-origin co-founder of AI startup Cursor that could be acquired for $60 billion by Elon Musk's SpaceX
Aman Sanger is emerging as one of the key Indian-origin figures in the global artificial intelligence boom. A co-founder of Cursor, he is part of the team behind one of the fastest-growing AI coding platforms in the world. Recently, SpaceX confirmed it had secured an option to acquire Cursor for around $60 billion, or alternatively, enter a $10 billion partnership. The development has brought renewed attention to its founders, including Sanger, whose journey reflects both global ambition and strong Indian roots.Aman Sanger is an Indian-origin entrepreneur and one of the four co-founders behind Cursor, alongside Michael Truell, Sualeh Asif, and Arvid Lunnemark. The team met while studying at the Massachusetts Institute of Technology, where they began exploring how artificial intelligence could transform the way software is written.Sanger grew up in the United States in a family with strong Indian roots. According to reports, his father, Arvind Sanger, is an alumnus of IIT Bombay and a hedge fund professional, while his mother, Shilpa Sanger, is an orthodontist and entrepreneur. This background helped shape his early interest in technology. He began coding at a young age and later studied computer science at MIT, where he co-founded Anysphere in 2022.He has played a key operational and strategic role in scaling Cursor from an early-stage startup into one of the fastest-growing AI developer platforms globally, despite maintaining a relatively low public profile.Cursor is an AI-powered coding platform designed to help developers write, edit, and understand code more efficiently. Unlike traditional tools focused on autocomplete, it functions as an intelligent collaborator that can analyse entire codebases and generate complex solutions.The platform has gained traction among developers and enterprises, positioning itself within a rapidly expanding category of AI coding tools. Its ability to reduce development time and improve productivity has made it a prominent player in the AI ecosystem.In 2026, SpaceX secured an agreement that includes an option to acquire Cursor for approximately $60 billion later in the year. Alternatively, the companies could pursue a $10 billion partnership focused on AI collaboration.This is not a completed acquisition. The arrangement allows both sides to work together while keeping the final decision open. If finalised, the deal would rank among the largest in the AI sector.The interest reflects a broader push by SpaceX into artificial intelligence. Advanced AI coding tools could accelerate software development across its operations, including rockets, satellites, and simulation systems. Integration with Musk's wider AI ambitions could also position Cursor within a larger ecosystem competing with major AI players.Since its founding, Cursor has grown rapidly, attracting widespread adoption among developers and enterprises. Its rise highlights a broader shift in the technology industry, where AI is becoming central to software development.The company's journey from a student project at MIT to a startup potentially valued in tens of billions reflects the speed at which AI-driven innovation is reshaping the global tech landscape.The potential $60 billion deal signals a deeper transformation in how software is built. AI is moving from a support tool to a core part of the coding process.For Aman Sanger, this marks a significant milestone. From his early exposure to technology in an Indian-origin family to co-founding a leading AI startup, he now stands at the centre of one of the most closely watched developments in the tech world. Whether or not the acquisition goes through, Cursor's rise and Sanger's role in it reflect the growing influence of artificial intelligence on the future of work.
Discord group says it accessed Claude Mythos by guessing location
The Anthropic AI model deemed a danger to cybersecurity may need to be more secure itself. An anonymous group of Discord users says it hacked its way into accessing Claude Mythos Preview, the new AI model Anthropic claims is too powerful for a public release. Anthropic says Claude Mythos "is capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser," and has granted access to the model to a select group of partners via an initiative called Project Glasswing. The AI company said this invite-only approach would let tech leaders "secure the world's most critical software." But it might need to pay more attention to its own software security. As Bloomberg reports, the Discord users didn't gain access through a sophisticated hack, but by guessing the online location for the model based on past Anthropic naming conventions -- as found in the recent data breach at Mercor, an AI startup, earlier this month. Once they identified where to access Claude Mythos, the group had to employ additional tactics. One member of the group already had privileged access as a worker at a third-party contractor for Anthropic, Bloomberg reports. The group was part of a private Discord channel that focuses on hunting information about unreleased models. A member of the group told Bloomberg that they were not using Claude Mythos for nefarious purposes, but for tasks like building simple websites. However, they also claimed to have access to even more unreleased Anthropic models. The group provided enough evidence to convince Bloomberg they had indeed breached Anthropic's security. Anthropic confirmed in a statement to Bloomberg it was aware of the claim and investigating. At this time, there is no indication that Claude Mythos has been breached by other unauthorized parties. Still, given that Anthropic described Claude Mythos as a paradigm-shifting security threat that could "reshape cybersecurity" as we know it, any unauthorized access is -- to say the least -- concerning. Want to learn more about getting the best out of your tech? Sign up for Mashable's Top Stories and Deals newsletters today.
SpaceX teams up with Cursor, secures option to acquire it for $60 billion
SpaceX said it has entered a deal with AI coding startup Cursor to develop a next-generation "coding and knowledge work AI." The deal also includes a provision that allows SpaceX to buy the startup for $60 billion later this year. "SpaceXAI and @cursor_ai are now working closely together to create the world's best coding and knowledge work AI," SpaceX said on X. "The combination of Cursor's leading product and distribution to expert software engineers with SpaceX's million H100 equivalent Colossus training supercomputer will allow us to build the world's most useful models." "Cursor has also given SpaceX the right to acquire Cursor later this year for $60 billion or pay $10 billion for our work together," the post noted. READ: Musk bought $1.4 billion worth SpaceX shares in 2025, report reveals (April 21, 2026) This comes shortly after Elon Musk's company confidentially filed for an IPO. Sources suggest the company may target a valuation of more than $1.75 trillion, making it one of the largest stock market listings ever. According to TechCrunch, investors seeking value in the IPO might see its engagement with Cursor as another way to extract value from the company. The TechCrunch report says this deal is unsurprising to those who follow the industry closely. Last week, it was reported that xAI would begin renting computing power from its data centers to Cursor, with the coding startup using tens of thousands of xAI chips to train its latest AI model. In February, two of Cursor's most senior engineering leaders, Andrew Milich and Jason Ginsberg, left the company to join xAI, where both report directly to Musk. SpaceX said this partnership was a project combining Cursor's "product and distribution to expert software engineers" with SpaceX's Colossus supercomputer, which the company claims has the equivalent compute power of a million Nvidia H100 chips. READ: SpaceX IPO: Elon Musk's rocket giant files confidentially (April 1, 2026) SpaceX also said that at some undisclosed point later this year, it will either pay Cursor $10 billion for its work or acquire the company for $60 billion. Another TechCrunch from last week revealed that Cursor was looking at a $50 billion valuation in an upcoming private fundraising round. This reveals a series of leaps -- Cursor was valued at just $2.5 billion in January 2025, climbed to $9 billion by May 2025, and was assigned a $29.3 billion post-money valuation when it closed on $2.3 billion in Series D funding in November of the same year. Cursor still uses and sells Claude and GPT models even as both firms now roll out their own coding tools. The SpaceX partnership might help the AI coding startup escape this awkward arrangement.
Microsoft bolsters cybersecurity with Anthropic AI
Microsoft has announced the integration of advanced artificial intelligence models, including Anthropic's Claude Mythos Preview, into its secure development framework. This initiative aims to enhance the detection and remediation of vulnerabilities from the earliest stages of software design. By incorporating these tools into its Security Development Lifecycle, the group intends to accelerate flaw identification and strengthen system robustness against escalating threats.
SpaceX Admits AI Data Centers in Space May Be a Really Terrible Idea
Can't-miss innovations from the bleeding edge of science and tech SpaceX CEO Elon Musk has been adamant that putting enormous AI data centers in Earth's orbit is a "no-brainer." At the World Economic Forum in Davos earlier this year, Musk claimed that the unconstrained solar power beyond our planet's surface makes it the "lowest-cost place to put AI," something "that will be true within two years, three at the latest." Yet many experts remain unconvinced that sending up to a million satellites, each bigger than the International Space Station, makes any sense, citing concerns over economic feasibility and physical limits. Some warn they could even cause an environmental catastrophe, with aging and failing hardware doomed to burn up in the Earth's atmosphere while releasing copious amounts of ozone-depleting chemicals. Even SpaceX itself isn't fully sold on the idea its mercurial CEO is pushing hard -- and prioritizing over his decades-old wish of sending humans to Mars. In excerpts of the company's pre-IPO filing obtained by Reuters, the company admitted that its "initiatives to develop orbital AI compute and in-orbit, lunar, and interplanetary industrialization are in early stages, involve significant technical complexity and unproven technologies, and may not achieve commercial viability." It's a significant tonal shift ahead of the company's blockbuster IPO, which is reportedly on track for later this year, at a record-breaking valuation of roughly $1.75 trillion. SpaceX acknowledged that sending sensitive AI chips into space may cause them to wear out much faster. Orbital data centers will operate "in the harsh and unpredictable environment of space, exposing them to a wide and unique range of space-related risks that could cause them to malfunction or fail." The company also pointed out its heavy reliance on its super heavy launch platform, Starship, to build out its enormous constellation of up to one million data centers. "Any failure or delay in the development of Starship at scale or in achieving the required launch cadence, reusability and capabilities thereof would delay or limit our ability to execute our growth strategy," the filing reads, as quoted by Reuters. SpaceX has yet to successfully send the rocket into space and safely land it in one piece. Previous test flights have resulted in massive mid-flight explosions. However, thanks to its enormous size and payload capacity, SpaceX could soon dramatically increase the amount of hardware it can launch into space, while theoretically lowering costs as well. In short, its plans for orbital data centers aren't just extremely ambitious; Musk has framed his entire space company on the endeavor months ahead of its long-awaited IPO. Yet glaring questions regarding the idea's financial, let alone physical, feasibility remain unanswered -- a reality that the company's engineers are clearly having trouble ignoring.
Microsoft to integrate Anthropic's Mythos into its security development program
April 22 - Microsoft said (MSFT.O), opens new tab on Wednesday it plans to embed advanced artificial intelligence models, including Anthropic's Claude Mythos Preview, into its secure coding framework, as the company steps up its cybersecurity capabilities. Incorporating the models into Microsoft's Security Development Lifecycle (SDL) will help identify vulnerabilities and develop fixes faster, early on in the cycle, the Windows maker said in a blog. Mythos, announced on April 7, has found "thousands" of major vulnerabilities in operating systems, web browsers and other software. Its capabilities to code at a high level have given it a potentially unprecedented ability to and devise ways to exploit them, experts said. Anthropic has said the current iteration, Claude Mythos Preview, will be first deployed to a select group of companies as part of Anthropic's "Project Glasswing," a controlled initiative under which major technology companies, including Microsoft, Amazon.com (AMZN.O), opens new tab and Apple (AAPL.O), opens new tab, can use it to search for cybersecurity vulnerabilities. Microsoft said it evaluated Mythos, using its own open-source benchmark for real-world detection engineering tasks, and the "results showed substantial improvements relative to prior models." U.S. President Donald Trump's administration, central bankers across the globe and industries are racing to get up to speed on Mythos and its ability to make complex cyberattacks both easier and quicker to crack. Reporting by Juby Babu in Mexico City; Editing by Shinjini Ganguli Our Standards: The Thomson Reuters Trust Principles., opens new tab
AI Security Breach: Anthropic Investigates Unauthorized Access to Powerful 'Claude Mythos' Model
AI startup Anthropic is investigating reports that a group of users gained unauthorized access to its Claude Mythos model, a powerful AI system that was deliberately released to only a small circle of trusted companies due to its advanced cybersecurity capabilities. The post AI Security Breach: Anthropic Investigates Unauthorized Access to Powerful 'Claude Mythos' Model appeared first on Breitbart.
Discord group says it accessed Anthropic's unreleased Claude Mythos
An anonymous group of Discord users says it hacked its way into accessing Claude Mythos Preview, the new AI model Anthropic claims is too powerful for a public release. Anthropic says Claude Mythos "is capable of identifying and then exploiting zero-day vulnerabilities in every major operating system and every major web browser," and has granted access to the model to a select group of partners via an initiative called Project Glasswing. The AI company said this invite-only approach would let tech leaders "secure the world's most critical software." But it might need to pay more attention to its own software security. As Bloomberg reports, the Discord users didn't gain access through a sophisticated hack, but by guessing the online location for the model based on past Anthropic naming conventions -- as found in the recent data breach at Mercor, an AI startup, earlier this month. Once they identified where to access Claude Mythos, the group had to employ additional tactics. One member of the group already had privileged access as a worker at a third-party contractor for Anthropic, Bloomberg reports. The group was part of a private Discord channel that focuses on hunting information about unreleased models. A member of the group told Bloomberg that they were not using Claude Mythos for nefarious purposes, but for tasks like building simple websites. However, they also claimed to have access to even more unreleased Anthropic models. The group provided enough evidence to convince Bloomberg they had indeed breached Anthropic's security. Anthropic confirmed in a statement to Bloomberg it was aware of the claim and investigating. At this time, there is no indication that Claude Mythos has been breached by other unauthorized parties. Still, given that Anthropic described Claude Mythos as a paradigm-shifting security threat that could "reshape cybersecurity" as we know it, any unauthorized access is -- to say the least -- concerning.
SpaceX Agrees to Potential $60B Deal to Acquire Cursor
Elon Musk's SpaceX has secured the option to buy AI coding startup Cursor for $60 million later this year. The agreement was made public in a post on Musk's X platform, which said the companies are "working closely together to create the world's best coding and knowledge work AI." The partnership will pair Cursor's code-writing software with SpaceX's Colossus supercomputer in Memphis, Tenn., which reportedly contains the equivalent of a million Nvidia H100 GPUs. As part of that arrangement, SpaceX will have the opportunity to acquire Cursor for $60 billion before the end of this year. Should it not do so, it will pay Cursor $10 billion for the work they have completed together -- a huge termination fee for a few months of partnership. Cursor was founded in 2022 and has since emerged as one of Silicon Valley's fastest-growing startups, achieving a valuation of $29.3 billion following a $2.3 billion Series D funding round in November of last year. While SpaceX was originally envisioned as a company that makes rockets and spacecraft, it has become increasingly involved in AI, as was illustrated by its $1.25 trillion merger with another of Musk's companies, xAI, earlier this year. With the firm now reported to be targeting an IPO, its ambitions appear to have evolved, with Musk apparently contemplating AI data centers in space following the success of the Starlink satellite program. "In the long term, space-based AI is obviously the only way to scale," he told employees when the SpaceX/xAI merger went through. How Cursor fits into this futuristic vision remains to be seen, but in the short term, the companies' potential integration will be viewed as an attempt to bolster xAI's ability to develop coding tools, given that the company has fallen well behind Claude Code and Codex in this area. That certainly appears to be how Cursor is approaching the deal, with CEO and co-founder Michael Truell posting on X: "Excited to partner with the SpaceX team to scale up Composer [Cursor's coding model]. A meaningful step on our path to build the best place to code with AI." A company blog post added more context, adding that attempts to advance its training efforts had been "bottlenecked by compute" and that leveraging Colossus would "dramatically scale up the intelligence of our models". SpaceX's post was as notable for what it omitted as what it revealed. The company has not disclosed which metrics will be used to determine whether to proceed with a deal, how the deal would be paid for, or whether it would go through before or after the mooted IPO.
Microsoft Integrates Anthropic's Mythos Into Security Development
Microsoft will embed Anthropic's Claude Mythos Preview into its Security Development Lifecycle (SDL) to detect and remediate vulnerabilities earlier in the software lifecycle, leveraging Mythos's advanced automated vulnerability discovery capabilities. Microsoft to Integrate Anthropic's Mythos AI into Secure Coding Framework Microsoft Embeds Anthropic's Claude Mythos Preview in Security Development Lifecycle April 22 - Microsoft said on Wednesday it plans to embed advanced artificial intelligence models, including Anthropic's Claude Mythos Preview, into its secure coding framework, as the company steps up its cybersecurity capabilities. Integration with Security Development Lifecycle (SDL) Incorporating the models into Microsoft's Security Development Lifecycle (SDL) will help identify vulnerabilities and develop fixes faster, early on in the cycle, the Windows maker said in a blog. Capabilities of Mythos AI Mythos, announced on April 7, has found "thousands" of major vulnerabilities in operating systems, web browsers and other software. Its capabilities to code at a high level have given it a potentially unprecedented ability to identify cybersecurity vulnerabilities and devise ways to exploit them, experts said. Deployment and Project Glasswing Anthropic has said the current iteration, Claude Mythos Preview, will be first deployed to a select group of companies as part of Anthropic's "Project Glasswing," a controlled initiative under which major technology companies, including Microsoft, Amazon.com and Apple, can use it to search for cybersecurity vulnerabilities. Evaluation and Benchmarking Microsoft said it evaluated Mythos, using its own open-source benchmark for real-world detection engineering tasks, and the "results showed substantial improvements relative to prior models." Industry and Global Response U.S. President Donald Trump's administration, central bankers across the globe and industries are racing to get up to speed on Mythos and its ability to make complex cyberattacks both easier and quicker to crack. (Reporting by Juby Babu in Mexico City; Editing by Shinjini Ganguli)
Microsoft to integrate Anthropic's Mythos into its security development program
April 22 - Microsoft said on Wednesday it plans to embed advanced artificial intelligence models, including Anthropic's Claude Mythos Preview, into its secure coding framework, as the company steps up its cybersecurity capabilities. Incorporating the models into Microsoft's Security Development Lifecycle (SDL) will help identify vulnerabilities and develop fixes faster, early on in the cycle, the Windows maker said in a blog. Mythos, announced on April 7, has found "thousands" of major vulnerabilities in operating systems, web browsers and other software. Its capabilities to code at a high level have given it a potentially unprecedented ability to identify cybersecurity vulnerabilities and devise ways to exploit them, experts said. Anthropic has said the current iteration, Claude Mythos Preview, will be first deployed to a select group of companies as part of Anthropic's "Project Glasswing," a controlled initiative under which major technology companies, including Microsoft, Amazon.com and Apple, can use it to search for cybersecurity vulnerabilities. Microsoft said it evaluated Mythos, using its own open-source benchmark for real-world detection engineering tasks, and the "results showed substantial improvements relative to prior models." U.S. President Donald Trump's administration, central bankers across the globe and industries are racing to get up to speed on Mythos and its ability to make complex cyberattacks both easier and quicker to crack. (Reporting by Juby Babu in Mexico City; Editing by Shinjini Ganguli)
SpaceX Warns of Potential Commercial Failure of Space Data Centers | ForkLog
SpaceX warns of potential commercial failure of space data centers. SpaceX has announced that Elon Musk's project to build data centers in space may not meet expectations. This was highlighted in the company's IPO filing, as reported by Reuters. "Our projects for developing orbital AI computing, as well as intra-orbital, lunar, and interplanetary industrialization, are in the early stages. They involve high technical complexity, use unproven technologies, and may prove to be commercially unviable," the document states. The aerospace firm is currently preparing for a stock market debut, which could potentially be the largest in history. SpaceX is targeting a valuation of about $1.75 trillion and plans to raise $75 billion in the coming months. In January, at the World Economic Forum in Davos, Musk described the creation of AI data centers in space as an "obvious solution" during a conversation with BlackRock CEO Larry Fink. According to him, orbit will become "the cheapest place to host AI capabilities" within two to three years. In February, SpaceX merged with AI startup xAI into a company valued at $1.25 trillion. At that time, the billionaire stated that the goal of the merger was to create data centers beyond the planet. "Current advancements in artificial intelligence rely on large terrestrial data centers, which require enormous amounts of energy and cooling. The global demand for electricity simply cannot be met with Earth-based solutions without harming people and the environment," he wrote. The idea is straightforward: satellites with AI chips can harness nearly constant solar energy in space, avoiding terrestrial limitations on space, energy, and cooling. However, turning this vision into reality is quite another challenge, as acknowledged by SpaceX. The IPO filing mentions that orbital data centers will operate in a "harsh and unpredictable space environment," where equipment may fail. SpaceX's advantage over competitors lies in its experience and technology. The company has already deployed the Starlink satellite network and is developing the fully reusable Starship rocket. According to Musk, this will reduce launch costs, making large-scale orbital infrastructure a reality. However, Starship itself carries risks. The rocket, designed to carry much heavier payloads than previous models, has already experienced failures during tests and has fallen behind schedule. SpaceX emphasized that further setbacks could limit its growth strategy.
