News & Updates
The latest news and updates from companies in the WLTH portfolio.
The Download: AI health tools and the Pentagon's Anthropic culture war
I've combed the internet to find you today's most fun/important/scary/fascinating stories about technology. 1 California has defied Trump to impose new AI regulations Governor Newsom signed off on the new standards yesterday. (Guardian) + Firms seeking state contracts will need extra safeguards. (Reuters $) + States are installing guardrails despite Trump's order to stop. (NYT $) + An AI regulation war is brewing in the US. (MIT Technology Review) 2 Experiments have verified quantum simulations for the first time It's a breakthrough for quantum computing applications. (Nature) + Which could one day help solve healthcare problems. (MIT Technology Review) 3 The new White House app is a security and privacy nightmare It extensively tracks users and relies on external code. (Gizmodo) + The new app promises "unparalleled access" to Trump. (CNET) + It also invites users to report people to ICE. (The Verge) 4 Big Tech's $635 billion AI spending faces an energy shock test The Middle East crisis is clouding prospects for growth. (Reuters $) + Here are three big unknowns about AI's energy burden. (MIT Technology Review) 5 Meta and Google have been accused of breaking child safety rules Australia suspects they flouted a social media ban. (Bloomberg $) + Indonesia is also investigating non-compliance. (Reuters $) 6 Nebius is building a $10 billion AI data center in Finland The company is rapidly expanding Europe's AI infrastructure. (CNBC)
'It's chaos' - Temporary traffic lights causing long delays in Ulverston
MOTORISTS are experiencing traffic delays on the A590 in Ulverston. There are currently temporary traffic lights in place close to the Booths roundabout. One motorist stuck in traffic in Ulverston described vehicles being 'bumper-to-bumper'. They added: "The traffic is barely moving. It's chaos." The AA is currently showing heavy traffic in the area. The AA is currently showing heavy traffic in Ulverston (Image: The AA) If possible, find alternative routes.
Easter holiday chaos ahead warning with major delays at Spanish airports
Ground handling workers at Barcelona, Madrid, Alicante, Malaga, Palma and Canaries airports are striking from Monday(Image: Jordan Lye via Getty Images) Strikes at one of Spain's most popular airports have caused long queues and several delays to services, with up to 800,000 passengers set to be impacted across the Easter holidays. Walkouts by ground‑handling staff plunged Palma Airport into chaos for a second day running, right as thousands of families began their Easter getaway. Frustrated passengers reported bottlenecks snaking through the departure halls, while airport sources confirmed that by midday on Tuesday (March 31), 12 flights had already been delayed. The strike, launched by the CCOO, UGT and USO unions, involves Groundforce workers and has been declared indefinite. Staff are downing tools in three daily waves, from 5-7am, 11am-5pm and 10pm-midnight, hitting all 12 Spanish airports where the company operates, including the busy Palama hub. Travellers had already been warned to prepare for delays, with the industrial action coinciding with the first major test of the EU's new Entry/Exit system, introduced late last year. Airport authorities expect around 800,000 passengers to pass through Palma over the Easter period, surpassing last year's figures.
Oracle Brings NVIDIA B300 GPUs and xAI Grok to Government Cloud Regions
Oracle is bringing NVIDIA's latest B300 GPUs and xAI's Grok models to its government cloud regions, giving federal agencies access to cutting-edge AI hardware previously limited to commercial customers. The expansion, announced March 31, 2026, targets agencies running demanding inference workloads under strict security requirements. The NVIDIA B300, powered by Blackwell Ultra architecture, represents the company's newest generation of AI accelerators. Oracle says the DGX B300 systems will help government customers train and run large language models more efficiently, with improved performance per watt compared to previous generations. Beyond raw compute, Oracle plans to offer xAI's Grok 4 and Grok 4 Fast models alongside NVIDIA Nemotron in government regions. The company positions these for specific use cases: Grok excels at data extraction, code generation, and text summarization, while Nemotron's throughput advantages suit agentic AI applications requiring fast response times. Federal customers currently have limited options for deploying frontier AI models in compliant environments. Most commercial AI services don't meet FedRAMP High or DISA Impact Level requirements that govern sensitive government workloads. The announcement lands alongside two related Oracle moves. The company also launched an isolated cloud specifically for defense contractors and unveiled a separate AI data platform targeting federal agencies -- all on the same day. Oracle's government cloud regions already hold FedRAMP High, DISA IL4, IL5, and IL6 authorizations. The company operates physically separated infrastructure from its commercial regions, a requirement for handling classified and controlled unclassified information. Current government cloud customers can already deploy AI applications using NVIDIA AI Enterprise software, including NIM microservices and NeMo training tools. The B300 expansion adds hardware muscle to that existing software stack. Agencies have struggled to keep pace with private sector AI adoption, largely due to compliance overhead and limited authorized options. Oracle's pitch: bring commercial-grade AI capabilities into environments that meet government security requirements without forcing agencies to build custom infrastructure. The company didn't specify exact availability dates, noting the infrastructure and model options depend on "service readiness and regional infrastructure availability." Government procurement cycles typically add months to deployment timelines regardless of vendor readiness. For Oracle, the expansion strengthens its position in a government cloud market where it competes against AWS GovCloud, Microsoft Azure Government, and Google Cloud's public sector offerings. The company has emphasized consistent pricing between commercial and government regions -- a differentiator when agencies compare total costs.
Tema Launches First Pure-Play Space ETF to Offer SpaceX Exposure | Weekly Voice
NEW YORK-(BUSINESS WIRE)-Tema ETFs ("Tema"), a leader in institutional-quality and actively managed exchange-traded funds, today announced the launch of the Tema Space Innovators ETF (NASA). NASA is the first pure-play ETF focused exclusively on the modern space economy to include 10% direct exposure to SpaceX, one of the world's most influential private companies. The global space economy is forecast to nearly triple by 2035,1 driven by falling launch costs, the rapid expansion of broadband
Holiday chaos as Spanish airports face strikes during busy Easter getaway period
Passengers reported bottlenecks across the departure halls(Image: Joaquin Ossorio-Castillo via Getty Images) Irish holidaymakers have faced long queues and early‑morning disruption following strikes at one of Spain's most popular holiday destinations. Strikes by ground‑handling staff entered their second day at Palma Airport, coinciding with the busy Easter getaway period. Passengers reported bottlenecks across the departure halls, with airport sources confirming that by midday on Tuesday (March 31), 12 flights had been delayed. The walkout, called by the CCOO, UGT and USO unions, involves Groundforce employees and has been declared indefinite. Staff are staging stoppages in three daily windows, from 5-7am, 11am-5pm and 10pm-midnight, affecting all 12 airports where the company operates, including Palma. Travellers had already been warned to prepare for delays, with the industrial action coinciding with the first major test of the EU's new Entry/Exit system, introduced late last year. Airport authorities expect around 800,000 passengers to pass through Palma over the Easter period, surpassing last year's figures.
OpenAI launches a Codex plugin that runs inside Anthropic's Claude Code
The plugin is part of a broader strategic shift: OpenAI plans to refocus on coding tools and enterprise customers, and to merge ChatGPT, Codex, and its Atlas browser into a single desktop app. OpenAI released a plugin that embeds its AI coding assistant Codex directly into Anthropic's Claude Code. The plugin offers three core features: a standard code review, a more aggressive "adversarial review" that specifically challenges weaknesses and design decisions, and the ability to hand off tasks to Codex in the background. It requires a ChatGPT subscription (including the free tier) or an OpenAI API key, plus Node.js version 18.18 or higher. The plugin doesn't install a separate program. Instead, it uses the Codex installation already on the machine, with the same settings and credentials. Users can also enable a "Review Gate" that prevents Claude Code from finalizing changes until Codex has reviewed them. Strategically, the plugin is a pragmatic move. Claude Code dominates the market right now. Rather than waiting for developers to switch to Codex, OpenAI is bringing its Codex model directly into their existing workflow. Every review triggered through the plugin runs on OpenAI's infrastructure and generates usage. At the same time, the plugin creates a direct touchpoint with the model. Developers get to experience Codex's capabilities in their day-to-day work without having to actively switch tools. That also applies to every update and every new model OpenAI ships in the coming weeks and months. The plugin fits into a broader strategic overhaul at OpenAI. According to a Wall Street Journal report, the company plans to refocus its resources on coding tools and enterprise customers. Fidji Simo, OpenAI's CEO of Applications, reportedly called the success of Anthropic's Claude Code and Cowork an internal "wake-up call." OpenAI also plans to merge ChatGPT, Codex, and its Atlas browser into a single desktop app. Greg Brockman is reportedly leading the transition. According to Simo, the previous strategy of launching too many products at once - including Sora, Atlas, and a hardware device - slowed the company down. The plugin is essentially a Trojan horse. The goal is to establish Codex as a useful addition to developers' workflows and gradually get them comfortable with OpenAI's ecosystem. It's available on GitHub at openai/codex-plugin-cc and has already collected more than 3,700 stars.
Airport strikes major update as Easter break chaos hits holiday spot
Spanish airport strikes show no sign of relenting as British tourists arriving and departing a popular holiday hotspot are met with significant delays and massive queues. Eye-witnesses at Palma Airport reported long lines at departure areas today, coinciding with the start of the busy Easter period. According to airport sources, twelve flights had already been delayed by midday this morning, consisting of eight departures and four arrivals. The industrial action, called by the CCOO, UGT, and USO unions, involves an indefinite strike by Groundforce ground staff that began on Monday. The walkouts are set to affect 12 airports where the company operates, including Palma, across three specific daily time slots. These slots are scheduled between 5am and 7am, 11am and 5pm, and 10pm to midnight Central European Time each day until further notice, reports The Express. British tourists have been warned to expect delays at the airport resulting from the industrial action. This will also be the first big test for the new European Union Entry and Exit system at Palma following its introduction at the end of last year. Around 800,000 people are expected to pass through the airport over the Easter period - a figure which is higher than last year.
Kraken: Krak Launches up to 2% Cashback, Its First Metal Debit Card, and In-App Concierge - Redefining Global Spending and Rewards
New offering includes tiered cashback of up to 2%, a premium Mastercard-powered Metal?? debit card, and Krak Concierge an in-app hotel and travel booking experience with exclusive rates and boosted rewards Krak, the global money app from Kraken, today announced the launch of its industry-leading debit cashback scheme of up to 2% for all customers, Krak's first Metal Mastercard debit card for eligible customers across the UK and EEA, alongside the launch of Krak Concierge a new in-app hotel and flight booking experience with boosted rewards up to 6%. This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20260331030305/en/ Together, these launches mark a major step in Krak's ambition to deliver high-value, real-world utility through a single global money app, combining everyday payments, premium rewards and lifestyle experiences in one seamless platform. "Your money should grow automatically. We're making rewards effortless with cashback that grows with you, setting a new standard for the global money app for travel and everyday spending," said Kamo Asatryan, Kraken's Chief Data Officer and Global Head of Consumer. A New, More Rewarding Cashback System Krak is introducing a new tiered cashback system designed to reward customers more as they deepen their relationship with the platform. The new structure offers progressively higher rewards based on a customer's 30-day average balance across Krak, Kraken and Kraken Pro. Customers can now earn up to 2% cashback on everyday spending, with rewards steadily increasing in line with the value of assets held on-platform. At the entry level, Krak customers holding up to £200 or €200 will earn 0.5% cashback, providing an immediate baseline of value from day one. From there, cashback rewards progressively scale as balances grow, unlocking higher tiers over time. At the top end, customers holding £50,000 or €50,000 or more can earn the full 2% cashback positioning Krak as one of the most rewarding debit card offerings available. By aligning rewards with customer balances, Krak delivers a simpler, more transparent and rewarding experience that grows alongside the customer turning everyday spending into a consistent source of value, without the need for subscriptions, credit, or complex conditions. Introducing Krak's first Metal Debit Card With over 60,000 Krak Coral and Black Mastercard debit cards issued in less than four months, Krak is now unveiling its first Metal Mastercard debit card to eligible customers across the UK and EEA. Crafted from precision-milled stainless steel, the Metal Card is powered by Payward's multi-asset infrastructure and designed for globally minded customers who expect more from everyday spending, whether at home or abroad. Powered by Mastercard's global payments network, the Krak Metal Card combines seamless payments and global acceptance, bringing users reliability and flexibility. Eligible Krak customers can spend effortlessly from more than 600 cryptocurrencies and fiat currencies through near-instant crypto to fiat conversion, with no FX or ATM withdrawal fees.* The Metal Card carries the same industry-leading uncapped cashback of up to 2% on purchases. Purchases are automatically funded according to the customer's preferred spending order, intelligently drawing from and combining multiple eligible balances where needed to cover the full amount, while excluding any assets the customer has opted to restrict. This gives Krak customers greater flexibility over how they store and spend their balances, removing the need for manual conversions, and reduces friction at the point of sale. "By deepening our partnership with Kraken with the launch of the Krak Metal Card, we're enabling more people to use crypto effortlessly in their everyday lives, connecting innovative digital solutions to the trusted global payments network they rely on," said Christian Rau, Senior Vice President, Global Digital Commercialization at Mastercard. Eligibility is available to customers who maintain an average balance of £50,000 or €50,000 across Krak, Kraken and Kraken Pro. Krak Concierge: Turn Travel Spending into Meaningful Rewards Krak also introduced Krak Concierge, a new in-app hotel and travel booking experience available to all Krak customers worldwide, designed to turn travel spending into meaningful rewards. Launched with streamlined hotel and flight booking services, Krak Concierge allows customers to earn boosted cashback of up to 6% on hotel bookings, while accessing exclusive, member-only rates across more than 2.2 million hotels and resorts worldwide, from affordable stays to five-star luxury destinations with savings of up to 60% compared to traditional booking platforms. Built in partnership with Entravel, the world's leading AI-powered crypto-native hotel booking platform, Krak Concierge brings premium travel and meaningful rewards directly into the Krak app. Customers can browse, book and pay seamlessly using hundreds of supported crypto and fiat balances in their Krak balances, or with Krak Cards. "Entravel is built on the principle of unlocking exclusive value through crypto integration," said Mathias Lundoe Nielsen, Founder and CEO of Entravel. "We are excited to empower the Kraken community with access to the world's best hotel prices and a seamless booking experience via Krak." Krak plans to expand its offering with additional premium merchants, exclusive perks and curated experiences in the months ahead, continuing to build a richer rewards ecosystem for its global customer base. What is Krak announcing today? Krak is upgrading its flat 1% cashback with a more rewarding tiered model based on 30-day average balances across Krak, Kraken and Kraken Pro, now offering up to 2% cashback opportunities. Krak is also launching a premium Metal debit card offering for eligible EEA and UK-based customers, powered by Mastercard. It is also introducing an in-app Concierge service, powered by Entravel, for booking hotel and flights, allowing customers to access member-only rates and earn boosted cashback rewards. About Kraken Founded in 2011, Kraken is one of the world's longest-standing and most secure crypto platforms globally. Kraken clients trade more than 600 digital assets, traditional assets such as U.S. futures and U.S.-listed stocks and ETFs, and 6 different national currencies, including GBP, EUR, USD, CAD, CHF, and AUD. Trusted by millions of institutions, professional traders and consumers, Kraken is one of the fastest, most liquid and performant trading platforms available. Kraken's suite of products and services includes the Kraken App, Kraken Pro, the Krak App, Kraken Institutional, Kraken's onchain offerings and the Ninja Trader retail trading platform. Across these offerings, clients can buy, sell, stake, earn rewards, send and receive assets, custody holdings, and access advanced trading, derivatives, and portfolio management tools. Kraken has set the industry standard for transparency and client trust, and it was the first crypto platform to conduct Proof of Reserves. It complies with regulations and laws applicable to its business, while actively protecting client privacy and maintaining the highest security standards. For more information about Kraken, please visit www.kraken.com. About Mastercard Mastercard powers economies and empowers people in 200+ countries and territories worldwide. Together with our customers, we're building a resilient economy where everyone can prosper. We support a wide range of digital payments choices, making transactions secure, simple, smart and accessible. Our technology and innovation, partnerships and networks combine to deliver a unique set of products and services that help people, businesses and governments realize their greatest potential. For more information about Mastercard, please visit www.mastercard.com. About Entravel Entravel is the world's leading crypto travel infrastructure provider, powering travel platforms for major crypto brands, fintechs and neobanks. Entravel offers private, members-only rates on over 2.2 million hotels, with discounts of up to 60% compared to mainstream booking platforms. *Note: A variable spread applies when spending across assets. Third-party ATM fees may apply. View source version on businesswire.com: https://www.businesswire.com/news/home/20260331030305/en/ Contacts: Press Contact Lauren Post [email protected] © 2026 Business Wire
ZachXBT: Kraken User Suspected Victim of Social Engineering Attack, Loses $18.2 Million - Lookonchain - Looking for smartmoney onchain
March 31 -- Per on-chain detective ZachXBT's monitoring, a Kraken user is suspected of falling victim to a social engineering attack, losing roughly $18.2 million. The attacker launched their operation roughly 45 minutes prior, using the SafePal wallet to transfer assets from the Ethereum network to Bitcoin's network via the decentralized cross-chain protocol THORChain.
Kraken User Drained in Suspected Social Engineering Heist
Blockchain investigator ZachXBT reported that an unknown Kraken user lost $18.2 million in a suspected social engineering scam, with stolen funds already moving across chains. The threat actor began bridging 878 ETH, worth roughly $1.8 million, from Ethereum (ETH) to Bitcoin (BTC) through THORChain using a SafePal wallet. ZachXBT shared the details on his Telegram channel. Another Social Engineering Victim Funneled Through THORChain The theft address tied to the incident is 0xC55149BbD560435a9FbEabFdcF9711cf928acA21 on Ethereum, with a corresponding BTC address of 1D8f8956EEFLXN28AHfioEx4ywVbxCz8KN. On-chain data from the THOR InfoBot confirmed the streaming swap was initiated roughly 45 minutes before the alert went public, adding credence to ZachXBT's claim. THORChain, a decentralized cross-chain liquidity protocol, does not require KYC verification. That feature has made it a repeated tool for laundering stolen funds. In January 2026, an attacker used the same protocol to move portions of $282 million in stolen BTC and Litecoin (LTC) after tricking a hardware wallet user into revealing their seed phrase. In that case, 818 BTC worth $78 million was swapped into ETH, XRP, and LTC through THORChain. The blockchain investigator also tracked a $330 million theft in 2025 and a $91 million loss in August of that year, both involving social engineering and cross-chain fund movement. The 878 ETH visible on-chain may represent only one tranche of the $18.2 million total being moved. Attackers in prior cases frequently split stolen assets across multiple transactions and chains simultaneously. Neither Kraken nor SafePal has publicly commented on the incident at the time of writing. The investigation remains in its early stages as on-chain trackers monitor further fund movements from the flagged addresses.
Paul Heyman Tells IShowSpeed To 'Relax' After WWE Chaos
WWE Hall of Famer Paul Heyman wants IShowSpeed to relax following his decision to join sides with The Vision. This week's Monday Night RAW episode saw a surprising tag team title change. Despite debuting a new theme song, Jey and Jimmy Uso of The Usos lost their tag-team titles to Logan Paul and Austin Theory of The Vision. Midway through the match, fans saw LA Knight trying to interfere and help The Usos, when IShowSpeed decided to step up. Speed took Paul's brass knuckles and proceeded to hit Knight in the face with it. In the end, Paul and Theory defeated The Usos to win the titles. The duo then proceeded to celebrate their title win with IShowSpeed on the ramp. Later in the show, backstage footage emerged which showed Speed contemplating his decision to join The Vision, when Heyman arrived from the shadows and asked him to "relax." "Hey, Relax. You are in good hands [Logan Paul & Austin Theory]. You are with the tag-team champions, you are with The Vision. They will protect [you]. Take him out to dinner with you. Please go celebrate." Heyman ended the segment by asking them to enjoy a celebratory dinner as the trio walked out. It is currently unknown if IShowSpeed has officially joined hands with Heyman and The Vision, or it was just a one-night scenario. Siding with Paul has not always ended well with Speed. At WrestleMania 40, Speed took a brutal RKO on the announcer's table during Paul's triple threat match against Kevin Owens and Randy Orton. Also in the show, fans witnessed another potential linkup between Gunther and Heyman as "The Ring General" saved Heyman from an ambush by Seth Rollins.
Anthropic's Claude source code leaked via a map file in npm registry, revealing hidden "Capybara" models and AI pet - Tech Startups
In what feels like the most on-brand slip of the year, Anthropic, the AI startup behind Claude, accidentally pushed its full Claude Code CLI source into the public npm registry -- again. The latest incident, tied to version 2.1.88, left behind a 57 MB source map file that effectively handed developers a clean, readable version of the entire TypeScript codebase. Within hours, the internet got to work. What emerged was a near-complete reconstruction of Claude Code's internal structure: more than 2,300 files, including the main interface, agent runtime, tooling system, and pieces that had never been announced. And yes, people noticed. "Claude code source code has been leaked via a map file in their npm registry," an X user shared. Code: https://pub-aea8527898604c1bbb12468b1581d95e.r2.dev/src.zip The reaction online landed somewhere between disbelief and amusement. One post summed up the mood: "They forgot to add 'make no mistakes' to the system prompt." That line stuck. The memes followed quickly. And for longtime observers, there was a sense of déjà vu. A similar source map issue surfaced in early 2025. The packages disappeared quietly back then. This time, the spotlight is brighter. To be clear, this isn't a leak of Claude's core intelligence. No model weights. No training data. What surfaced is the client-side layer -- the CLI tool developers install and run locally. Still, it's a detailed look at how Anthropic structures its agent workflows, tool-calling logic, and interface decisions. And there's plenty to look at. Before anyone panics: this is not the model weights. No secret sauce, no Claude LLM brain, no training data. It's 100% the frontend/CLI JavaScript/TypeScript client -- the terminal agent you npm install and run locally. But it's still a massive leak. The full production codebase includes: Community archaeologists have already unearthed some gems: Developers combing through the code uncovered references to a new model family named "Capybara," split into multiple tiers. There are traces of experimental features that never made it into public releases. One of the more unexpected finds is a hidden command that points to a Tamagotchi-style AI companion system, complete with traits, rarity levels, and visual elements. Then there's telemetry. The code shows that Claude Code tracks user behavior in ways that are more specific than many expected. It logs frustration signals, such as when users swear, and patterns, such as repeated "continue" prompts. That data appears to be routed through Datadog, along with session metadata such as model usage and environment details. The code includes safeguards to prevent the transmission of user code or file paths, and users can disable telemetry entirely via environment variables. For many developers, the more serious takeaway isn't the features -- it's the visibility into the system itself. The permissions model is now out in the open. The logic behind tool approvals, file access, and execution boundaries can be read line by line. The full system prompt is also exposed, giving a clear picture of how Anthropic guides Claude's behavior and enforces limits. For anyone interested in stress-testing or bypassing those controls, this is a detailed map. There's no sign that user data has been compromised. The infrastructure details revealed -- session handling, authentication flows, feature flags -- offer insight, not direct access. Still, this level of transparency changes the conversation around how these tools are secured in practice. What makes this moment stand out isn't just the mistake. It's what happens next. Developers have already started forking the code. Some are stripping it down. Others are experimenting with lighter, faster versions. There's talk of community-built alternatives that take advantage of what's now visible. That kind of momentum tends to move quickly. Anthropic hasn't issued a public response yet. The company will likely rotate exposed keys and tighten its release process. That part is routine. The harder question is how a slip like this happens more than once in a product used by developers who expect precision. For now, the story is still unfolding. The code is out there. The forks are multiplying. And somewhere, someone is double-checking their build pipeline before the next publish goes live.
Anthropic's Claude Code Source Code Reportedly Leaked Via Their npm Registry
Anthropic's proprietary Claude Code CLI tool has had its full TypeScript source code inadvertently exposed through a misconfigured npm package, after a security researcher discovered a leaked file referencing the unobfuscated codebase stored on Anthropic's own cloud infrastructure. On March 31, 2026, researcher Chaofan Shou made the disclosure public, posting: "Claude code source code has been leaked via a map file in their npm registry!" The published npm package reportedly contained a source map () file that referenced the complete, unminified TypeScript source, which was directly downloadable as a ZIP archive from Anthropic's own R2 cloud storage bucket. The original unmodified source has since been preserved and mirrored in a public GitHub repository under the backup branch . What Was Exposed The leaked codebase represents the entirety of Claude Code's directory, approximately 1,900 files and over 512,000 lines of code written in strict TypeScript, using the Bun runtime and a React + Ink terminal UI framework. The disclosure is substantial in scope, touching every critical subsystem of the CLI tool. Key files confirmed in the leak include (~46,000 lines), which drives the core LLM API engine, including streaming, tool loops, and token tracking; (~29,000 lines), defining all agent tool types and permission schemas; and (~25,000 lines), which registers and executes the tool's slash commands. The exposed architecture details approximately 40 agent tools, including , , , and for sub-agent spawning as well as roughly 85 slash commands spanning Git workflows, code review, memory management, and multi-agent orchestration. The leak also reveals internal feature flags such as , , , and , indicating product features not yet publicly released. How Source Maps Cause Leaks Source map files () are standard developer tools designed to map compiled or minified JavaScript back to its original source for debugging purposes. When inadvertently bundled into production npm releases, however, they expose proprietary source code to anyone who knows where to look effectively bypassing obfuscation entirely. This is not the first time such an error has affected Anthropic; a similar source map exposure was reportedly patched in early 2025. The breach raises serious intellectual property concerns for Anthropic, as the exposed code covers internal API client logic, OAuth 2.0 authentication flows, permission enforcement, multi-agent coordination, and even undisclosed feature pipelines. At the time of writing, Anthropic has not issued a public statement regarding the incident. Organizations integrating Claude Code into their development workflows should monitor Anthropic's official security advisories. Developers are urged to review the official npm registry for patched releases and avoid third-party mirrors of the leaked source. Note: This article is based on publicly available disclosures. Cybersecurity News does not host or distribute the leaked source code.
Bridgeport Motorsports Park owners apologize for chaos around festival
Capt. Howard Bennett, a Camden firefighter, was laid to rest Feb. 11 and honored with a celebration of life. LOGAN TWP. -- The owners of Bridgeport Motorsports Park have issued a lengthy apology for massive traffic-related problems that developed around a car and music festival held March 29 at their Floodgate Road facility. Logan Township police issued a statement March 30 that conditions around the "Import Export" festival got so out of hand before 11 a.m. Sunday that its mutual assistance partners across Gloucester County came in to help with traffic and crowd control. Doug and Brittany Rose took over in 2019 as owners and operators of the park, known in the automotive racing community as the "Kingdom of Speed." They posted an apology and explanation on the park Facebook page early Monday afternoon. "First and foremost, my wife and I would like to apologize to the community for yesterday's inconveniences," the post reads. "I do not like to make excuses, and I do not want this to sound like an excuse. "I would like to express to our neighbors, once we realized this was not what we expected as a facility, we, along with authorities diligently began to dissolve the event without causing things to escalate," the Roses stated. Management could not be reached by phone or email on Monday. Logan police say Sunday's event lacked local approval. The township also did not receive notice it was happening, police say. "The area rapidly became inundated with motorists from as far away as Massachusetts and Virginia," police stated. "The park quickly exceeded its capacity and an estimated 25,000 individuals parked wherever they chose, exited their vehicles, and began walking to the venue. The park website does include a schedule of 2026 events. The Import Export event is part of that list, including a provision for April 12 in case of rain. According to Logan police, motor vehicle violations reported included speeding, unsafe passing, racing, and passengers riding atop vehicles. In addition, they say, reports came in of public intoxication, public urination, lewdness, disorderly conduct, littering, and fights. An emergency alert message was sent out with the help of the county Office of Emergency Management. The alert was lifted around 6:44 p.m. Sunday. Police say one male was charged with disorderly conduct. He was released pending a hearing in Municipal Court. The Roses stated that traffic was the basic problem. And after access to Floodgate Road was closed to vehicles, people parked and walked to the park. "At that time, enough was enough and we decided to begin the process of shutting down the event, again with the mindset not to cause anything to escalate," the owners posted. "Once the event stopped and everyone was made aware is when the alerts went out over the phones asking everyone to disperse." Joe Smith is a N.E. Philly native transplanted to South Jersey almost four decades ago, concentrating on housing, politics, real estate, business, and development. He is a former editor and current senior staff writer for The Daily Journal in Vineland, Courier-Post in Cherry Hill, and the Burlington County Times. Have a tip? Support local journalism with a subscription.
Anthropic's Runaway Demand Is Breaking Its Own Infrastructure -- and Reshaping the AI Arms Race
Anthropic has a problem most startups would kill for. Demand for its Claude AI models is surging so fast that the company can't get enough servers to keep up. The result: a supply crunch that is forcing one of the most closely watched artificial intelligence companies to ration access, rethink its infrastructure strategy, and scramble for computing power at a pace that even its own leadership didn't fully anticipate. The server shortage, first reported by The Information, has become acute enough that Anthropic has at times struggled to reliably serve its enterprise customers -- the very clients paying top dollar for access to Claude's capabilities. Internal discussions have centered on how to allocate scarce GPU capacity across competing priorities: the consumer-facing chatbot, the API that powers thousands of developer applications, and the massive training runs needed to build next-generation models. Every allocation decision carries a trade-off. More capacity for inference means less for training. More for enterprise means less for the free tier. It's a tension that sits at the heart of every scaling AI company right now, but Anthropic's version is particularly intense because of how quickly its commercial traction has accelerated. The company's annualized revenue reportedly crossed $2 billion earlier this year, a figure that has roughly doubled in a matter of months. Claude's popularity among developers and businesses -- particularly its Claude 3.5 Sonnet model, which has earned a reputation for strong coding and reasoning performance -- has driven usage volumes that outstripped the infrastructure Anthropic had provisioned. According to The Information, the company has been in active negotiations to secure additional server capacity from cloud providers, including Amazon Web Services and Google Cloud, both of which are also investors in Anthropic. That dual relationship -- customer and investor -- adds a layer of complexity. AWS, which has committed up to $4 billion in Anthropic, hosts Claude through its Bedrock platform. Google, which has invested roughly $2 billion, offers Claude through Vertex AI. Both cloud giants benefit from Anthropic's success driving workloads onto their platforms. But both also have their own AI models to promote. Google has Gemini. Amazon is building its own Nova models. The incentive to prioritize Anthropic's capacity requests isn't always perfectly aligned with their own competitive interests. So Anthropic finds itself in an unusual bind: deeply dependent on its rivals for the very infrastructure it needs to compete against them. The GPU Scramble and What It Signals About AI's Real Bottleneck The broader context here matters. The global supply of high-end AI chips -- overwhelmingly Nvidia's H100 and the newer H200 and Blackwell GPUs -- remains constrained despite Nvidia's record-breaking production ramp. Every major AI lab, every hyperscaler, and an expanding list of sovereign wealth funds and government-backed initiatives are competing for the same silicon. Anthropic isn't alone in feeling the squeeze. But its particular growth trajectory has made the mismatch between demand and supply especially visible. Anthropic CEO Dario Amodei has spoken publicly about the capital intensity of the AI race. In interviews, he's described a world where frontier model training runs could cost $5 billion to $10 billion within the next few years. The company closed a $2 billion funding round led by Lightspeed Venture Partners in early 2025, bringing its total raised to roughly $13.7 billion. It's an extraordinary sum. And yet the server crunch suggests it may not be enough to stay ahead of the curve. Part of the issue is timing. Provisioning tens of thousands of GPUs doesn't happen overnight. Data center buildouts take months or years. Even when chips are available, the physical infrastructure -- power, cooling, networking -- has to be in place first. Anthropic's demand curve has moved faster than the physical world can accommodate. There's also the question of efficiency. Anthropic has invested heavily in research to make its models more compute-efficient, and Claude 3.5 Sonnet was widely praised for delivering strong performance at lower inference costs than some competitors. But efficiency gains can be a double-edged sword: lower per-query costs tend to stimulate even more usage, a dynamic economists call the Jevons paradox. Make it cheaper to run, and people run it more. The company has reportedly explored several mitigation strategies. Rate limiting for free-tier users. Priority queuing for paying enterprise customers. Dynamic load balancing across cloud providers. And, critically, conversations about building or leasing its own dedicated data center capacity rather than relying entirely on AWS and Google Cloud. That last option would represent a significant strategic shift -- moving Anthropic closer to the vertically integrated model that OpenAI has pursued through its partnership with Microsoft and its involvement in the Stargate project. OpenAI's own infrastructure trajectory offers a useful comparison. Microsoft has committed tens of billions of dollars to building out data center capacity specifically for OpenAI's workloads, a level of dedicated investment that gives OpenAI a structural advantage in raw compute availability. Anthropic doesn't have an equivalent arrangement. Its cloud partnerships are real but more transactional. The server crunch is, in part, a consequence of that difference. The competitive implications are significant. If Anthropic can't serve enterprise customers reliably, those customers will look elsewhere. Large organizations evaluating AI providers care about uptime, latency, and consistent availability just as much as they care about model quality. A technically superior model that's frequently unavailable or throttled loses deals to an adequate model that's always on. And the enterprise AI market is moving fast. According to recent reporting from Reuters, corporate AI spending is accelerating across financial services, healthcare, and software development, with companies increasingly splitting workloads across multiple providers to avoid exactly the kind of single-provider dependency risk that Anthropic's crunch illustrates. Multi-model strategies are becoming the norm, not the exception. Anthropic's leadership is aware of the stakes. The company has been hiring aggressively for infrastructure and operations roles, and its job postings in recent months have emphasized experience with large-scale distributed systems, GPU cluster management, and cloud capacity planning. These aren't research hires. They're the kind of operational talent you bring in when you're trying to build and manage infrastructure at hyperscaler scale. There's an irony in all of this. Anthropic was founded in 2021 by former OpenAI executives, including Dario and Daniela Amodei, with a stated mission to build AI that is safe and interpretable. The company's identity has been rooted in careful, research-driven development -- a deliberate contrast to what its founders saw as OpenAI's increasingly aggressive commercialization. But commercial success has a way of imposing its own logic. When your product is this popular, the pressure to scale infrastructure becomes existential, not optional. The safety mission hasn't been abandoned. Anthropic continues to publish research on constitutional AI, model alignment, and interpretability. But the day-to-day operational reality increasingly looks like that of any fast-scaling technology company: fighting for capacity, managing customer expectations, and trying to build infrastructure fast enough to keep up with a demand curve that keeps steepening. For the broader AI industry, Anthropic's server crunch is a leading indicator. The companies building frontier models are entering a phase where the binding constraint isn't algorithmic innovation -- it's physical infrastructure. Chips, power, cooling, data center space. The winners of the next phase of the AI race may not be the ones with the best models. They may be the ones who can actually run them at scale. That's a sobering thought for a field that has spent the last three years focused almost entirely on model architecture and training techniques. The hard part, it turns out, might be the plumbing.
Spain airport strikes major update as Easter holiday chaos hits Brit hotspot
According to airport sources, twelve of the flights were delayed by midday this morning, eight of which were departures and four were arrivals. The CCOO, UGT and USO unions have called an indefinite strike by Groundforce ground staff, which began on Monday. The action will take place across three different time slots and is set to affect the 12 airports where the company operates, including Palma. The timeslots are between 5am and 7am, 11am and 5pm and 10pm to midnight Central European Time, each day indefinitely.
SpaceX loses contact with one of its Starlink satellites
SpaceX has lost contact with Starlink satellite 34343 after it suffered an unspecified anomaly on March 29 while it was in orbit, the company has announced on X. The event happened while the satellite was approximately 348 miles above our planet. Since that is a relatively low altitude, SpaceX's analysis showed that the remains of the satellite pose no risk to the International Space Station or the upcoming launch of the Artemis II mission. It also won't affect the company's Transporter-16 mission, which launched with small satellites from its clients on March 30. In its statement, SpaceX also said that it will monitor any trackable debris, indicating that the satellite is no longer in one piece. LeoLabs, an American company tracking satellites in Low Earth Orbit, said it detected a "fragment creation event" involving Starlink 34343 on March 29. It also mentioned that this event is similar to another incident that happened on December 17, 2025. SpaceX had lost Starlink satellites to events like geomagnetic storms in the past, but it doesn't seem like these two recent incidents were caused by external factors. The company has yet to announce what led to the anomalies, but LeoLabs believes that both of them were "likely caused by an internal energetic source rather than a collision with space debris or another object." At the moment, the Starlink team is still working on determining the root cause of the incidents. SpaceX said that once it has come to a conclusion, the company will "rapidly implement any necessary corrective actions."
Scotland drives UK satellite ambitions with SpaceX Transporter-16 mission launch
Three satellites built in Glasgow have successfully launched aboard a SpaceX Transporter-16 mission Developed by Spire Global and AAC Clyde Space, the SpaceX Transporter-16 mission shows Scotland's growing role as a hub for advanced space technology and manufacturing. With funding from the UK Space Agency and delivered through the European Space Agency Pioneer Programme, the satellites demonstrate cutting-edge capabilities in laser communications, spacecraft production, and constellation operations. The programme is part of the Advanced Research in Telecommunications Systems initiative, designed to accelerate innovation in satellite communications. One of the satellites will test optical inter-satellite link (ISL) technology. This approach uses lasers instead of traditional radio frequencies to transfer data between satellites. Developed by Spire Global's UK division, the payload is designed for a compact nanosatellite platform while delivering high-speed data transfer. This innovation could significantly reduce latency, allowing for real-time data delivery across global satellite networks. Industries such as aviation, maritime operations, weather forecasting, and space weather monitoring are expected to benefit from faster, more reliable connectivity. Alongside the ISL mission, two additional satellites developed by AAC Clyde Space take the next steps in high-volume satellite production. Built under the xSPANCION project, the satellites showcase the UK's ability to manufacture spacecraft at scale while maintaining cost efficiency. The project brought together several partners, including the University of Strathclyde and Satellite Applications Catapult, reinforcing collaboration across academia and industry. These satellites will become part of VIREON™, a constellation designed to deliver actionable insights for agriculture, forestry, and environmental management. The launch of these three satellites aboard the SpaceX Transporter-16 mission shows a bigger national strategy to prioritise satellite communications as an area of growth. The UK government has committed over £600 million to research and development in this field, recognising its importance for both civilian infrastructure and defence applications. Satellite communications play an important role in providing broadband to remote regions, supporting emergency response systems, and ensuring secure military connectivity. As demand for global connectivity continues to grow, investments in this sector are expected to drive job creation and technological leadership. The success of the SpaceX Transporter-16 mission helps cement Glasgow's reputation as a leading centre for small-satellite manufacturing in Europe. The city has become a focal point for innovation, with companies and research institutions working together to push the boundaries of what small satellites can achieve. With additional funding opportunities available through programmes such as the UK Space Agency's Connectivity in Low Earth Orbit initiative, more companies are expected to enter the market and contribute to the UK's expanding space ecosystem.
Anthropic admits Claude Code quotas running out too fast
Unexpected quota drain prompts complaints, breaks automated workflows Users of Claude Code, Anthropic's AI-powered coding assistant, are experiencing high token usage and early quota exhaustion, disrupting their work. Anthropic has acknowledged the issue, stating that "people are hitting usage limits in Claude Code way faster than expected. We're actively investigating... it's the top priority for the team." A user on the Claude Pro subscription ($200 annually) said on the company's Discord forum that "it's maxed out every Monday and resets at Saturday and it's been like that for a couple of weeks... out of 30 days I get to use Claude 12." The Anthropic forum on Reddit is buzzing with complaints. "I used up Max 5 in 1 hour of working, before I could work 8 hours," said one developer today. The Max 5 plan costs $100 per month. There are several possible factors in the change. Last week, Anthropic said it was reducing quotas during peak hours, a change that engineer Thariq Shihipar said would affect around 7 percent of users, while also claiming that "we've landed a lot of efficiency wins to offset this." March 28 was also the last day of a Claude promotion that doubled usage limits outside a six-hour peak window. A third factor is that Claude Code may have bugs that increase token usage. A user claimed that after reverse engineering the Claude Code binary, they "found two independent bugs that cause prompt cache to break, silently inflating costs by 10-20x." Some users confirmed that downgrading to an older version helped. "Downgrading to 2.1.34 made a very noticeable difference," said one. The documentation on prompt caching says that the cache "significantly reduces processing time and costs for repetitive tasks or prompts with consistent elements." That said, the cache has only a five-minute lifetime, which means stopping for a short break, or not using Claude Code for a few minutes, results in higher costs on resumption. Developers can upgrade the cache lifetime to one hour but "1-hour cache write tokens are 2 times the base input tokens price," the documentation states. A cache read token is 0.1 times the base price, so this is a key area for optimization. Anthropic does not state the exact usage limits for its plans. For example, the Pro plan promises only "at least five times the usage per session compared to our free service." The Standard Team plan promises "1.25x more usage per session than the Pro plan." This makes it hard for developers to know what their usage limits are, other than by examining their dashboard showing how much quota they have consumed. Problems like this are not unusual. Earlier this month, users of Google Antigravity were protesting about similar issues. Bugs aside, what we are seeing is an implicit negotiation between users and providers over what is an acceptable pricing and usage model for AI development. Users want to control costs and providers need to make a profit. There is also a disconnect between vendor marketing that urges developers to insert AI into every process, including in some cases automated workflows, and a quota system that can cause AI tools to stop responding. "For folks running Claude Code in automated workflows: rate-limit errors need to be caught explicitly - they look like generic failures and will silently trigger retries. One session in a loop can drain your daily budget in minutes," observed one user. ®
