News & Updates
The latest news and updates from companies in the WLTH portfolio.
Pentagon still considers Anthropic a supply chain risk, official says
His comments came a day after Commerce Secretary Howard Lutnick said Anthropic was "back on the right side," adding, "we trust Anthropic" during an interview with Axios WASHINGTON, D.C.: Anthropic remains classified as a "Supply Chain Risk" by the U.S. Defense Department and the defense industry, a senior Pentagon official said, following suggestions that relations between the AI company and the Trump administration were improving. Emil Michael, under secretary of defense for research and engineering, made the remarks in a post on X. His comments came a day after Commerce Secretary Howard Lutnick said Anthropic was "back on the right side," adding, "we trust Anthropic" during an interview with Axios. A federal judge ruled in Anthropic's favor on Aug. 27, determining that the Pentagon had illegally punished the AI company for criticizing the government and that the Pentagon's decision was "illegal and baseless." The Defense Department did not respond to a request for comment.
Pentagon still considers Anthropic a supply chain risk, official says
His comments came a day after Commerce Secretary Howard Lutnick said Anthropic was "back on the right side," adding, "we trust Anthropic" during an interview with Axios WASHINGTON, D.C.: Anthropic remains classified as a "Supply Chain Risk" by the U.S. Defense Department and the defense industry, a senior Pentagon official said, following suggestions that relations between the AI company and the Trump administration were improving. Emil Michael, under secretary of defense for research and engineering, made the remarks in a post on X. His comments came a day after Commerce Secretary Howard Lutnick said Anthropic was "back on the right side," adding, "we trust Anthropic" during an interview with Axios. A federal judge ruled in Anthropic's favor on Aug. 27, determining that the Pentagon had illegally punished the AI company for criticizing the government and that the Pentagon's decision was "illegal and baseless." The Defense Department did not respond to a request for comment.
US government now 'trusts' Anthropic, the company that it banned months ago; Howard Lutnick says: They are back on ...
In what can be called a major diplomatic reversal following months of bitter friction over artificial intelligence (AI) safety protocols and national defense boundaries, the Trump administration officially trusts Anthropic, Commerce Secretary Howard Lutnick confirmed this week. Speaking in an interview with Axios, Lutnick responded to a direct question whether he places confidence in Anthropic CEO Dario Amodei."We trust Anthropic. They've done what we asked. They're back on the right side. So the answer is: Yes," Lutnick responded. The remarks signal a major de-icing of ties that earlier this year resulted in sweeping export restrictions targeting Anthropic's most capable foundation models. As per the report, personal friction between top officials and corporate leaders had also compounded those clashes, driving the firm away from executive favour.The report says that the turnaround in relations arrived after Anthropic co-founder Tom Brown assumed a central role in managing the startup's relationship with Washington. Lutnick showcased the improved alliance by personally introducing Brown before a global audience of dignitaries at the G20 Innovation Ministerial, inviting the co-founder to tell his story on the international stage, the report noted.In behind-the-scenes diplomacy, Brown proved critical in mending fences across official channels, reportedly conducting extensive bilateral discussions with Lutnick and National Cyber Director Sean Cairncross.Taking the ministerial stage, Brown openly commended a Truth Social post shared by President Donald Trump praising data center expansion. Brown highlighted that advanced facilities generate tax revenues, create employment and are engineered to inject new electrical capacity into the grid.The shift comes amid shifting courtroom battles over prior regulatory crackdowns. A federal judge recently struck down the Pentagon's blacklisting of Anthropic as a supply-chain hazard, determining that the punitive designation infringed upon the technology company's constitutional protections.Anthropic is continuing to litigate an additional Pentagon designation invoked under a different statute before the US Court of Appeals for the District of Columbia Circuit.
Anthropic back on 'Trump's side': Report
US Commerce Secretary Howard Lutnick said AI giant Anthropic is "back on the right side" with the Trump administration, Axios reported on Wednesday citing an interview. US Defense Secretary Pete Hegseth had blocked Anthropic from certain military contracts because the company refused to allow the US to use its Claude AI models for domestic surveillance or autonomous weapons. Anthropic sued in California court. On 27 August, a US judge sided with Anthropic in the company's fight with the American military over AI safety on the battlefield. Anthropic and the Commerce Department did not immediately respond to requests for comment from Reuters. Anthropic co-founder Tom Brown appeared on stage at a tech-focused G20 event in North Carolina on Wednesday morning. Brown stressed the importance of data centers at the gathering, urging the gathered officials to grant permits to build them in their countries.
Anthropic's Content Checker Tool Is Here, With One Big Catch - CNET
Omar Gallaga has covered technology, digital culture and other topics for outlets including CNET, NPR, WIRED, Texas Monthly, MSNBC, Consumer Reports, The Washington Post, the Los Angeles Times, The Atlantic and the Austin American-Statesman, where he was a longtime tech reporter, editor and podcaster.... Read full bio Just a couple of weeks after Anthropic announced it would roll out digital watermarks for text generated with its Claude AI tools, the company has introduced a page where you can check files to see if Claude helped make them. The page, at claude.com/check-content, lets you upload files with a quick spot check and will confirm, "This file shows signs it was processed by Claude," if it finds content credentials embedded in the file. But if you're hoping to prove that someone used Claude for a term paper or want to see if someone's texts weren't actually written by the human you're talking to, you're going to need to wait. The checker currently works with only a limited number of file types, and none of them are text formats. Instead, the checker accepts multimedia files, including JPEGs, GIFs, MP3s and MOVs -- basically image, audio or video files. A spokesperson for Anthropic told CNET that the tool is specifically to check for Claude's use through an industry standard called C2PA; a broader AI detection tool is in the works, but not yet available. "Text checking is via our Detection API (separate tool) and in private preview, and we hope to share more about it in the months ahead," the spokesperson said. This aligns with a note on the check content tool reads, "To detect watermarks embedded in text we have a Detection API which is currently in private preview to eligible organizations as required under EU law." Anthropic has said that its watermarking of text that Claude produces won't affect the quality, speed or cost of outputs and that these watermarks won't be visible or affect how text functions or appears. But they will remain with text from Claude even if it's copied and pasted from one place to another. Anthropic has a sign-up form for businesses or organizations that want to be added to the list for access to the watermark detector. Anthropic said in its email to CNET that if the C2PA standard expands to include additional file formats, its checker tool will likely adopt them as well. How does Anthropic's new checker work? The free web-based tool has a drag-and-drop option, or you can click to add a file from your device or Google Drive on mobile. According to the website, the file you choose stays on your device or file service, and the checker only reads the file's credentials, not the file itself. "Your file is never stored or used for any other purpose," the page says. In CNET's testing of the tool, the detector correctly flagged an image file that was converted in Claude from a PowerPoint presentation. For files that the detector said were not shown to be processed by Claude, a message is shown that reads, "We couldn't find any signs that this file was processed by Claude. That doesn't rule it out. Signals like this can be removed when a file is edited, converted, or compressed."
Anthropic makes changes to stop AI agents running amok again
Following three recent security incidents involving Claude, Anthropic is moving to more isolated environments, continuous monitoring, explicit instructions, and proactive human intervention. endif; ?> Learning from the OpenAI-Hugging Face fiasco, as well as from recent revelations about its own model, Anthropic is revamping its security and alignment practices. The company has established controls that flag when a model attempts to break out of a sandbox or successfully accesses the live internet, cordoned off its highest-risk test environments, and proposed a set of safety standards for its external testing partners, such as giving AI agents explicit instructions like "you should not access the internet." Anthropic conceded that three recent security incidents involving Claude reflect a "failure of operational security," and also reveal issues with model reasoning capabilities and "recklessness." Recent events "stressed that the urgency of improving our cybersecurity defenses is even higher than we previously believed," the company noted.
Anthropic's standard could let AI agents control different machines
* Anthropic wants AI agents to control programmable machines through common software rules * Model Hardware Standard (MHS) can connect microscopes, robotic arms, and liquid handling equipment * MHS allows different laboratory devices to communicate through standardized software drivers Anthropic opened a preview of its Model Hardware Standard (MHS), designed to let AI agents control programmable machines through shared rules. The proposed system is intended for laboratories, factories, and workplaces where different instruments must work together without custom software links. MHS uses software drivers that translate between computer systems and individual machines, allowing equipment with different interfaces to communicate consistently. A common system for physical machines Its drivers rely on basic commands for reading measurements or changing settings, giving agents a method for controlling connected equipment. The system records machine details that might otherwise remain inside manuals or depend on knowledge held by experienced laboratory staff. Users can enter such information through natural language, after which MHS creates reference material describing capabilities, adjustments and safety restrictions. Agents can then discover compatible equipment across networks without requiring separate software bridges for every machine they need to operate. The standard supports microscopes, liquid handlers, robotic arms and other equipment that offers some form of programmable interface for agents. The company claims integration that once required weeks or months could instead take hours or minutes when equipment supports MHS. According to Anthropic, Claude was tested with physical experiments, including laser alignment supported by camera-based observation during early physical trials. The model successfully adjusted the laser, checked the image, and repeated the process while assessing each change carefully. The design also lets agents combine commands from several devices, creating automated sequences for work that might otherwise require coordination. Physical limits remain a concern Anthropic said Claude interacts with experiments and hardware in an "exploratory manner, much as a scientist would," during physical testing. That model can then coordinate several instruments through one interface, rather than requiring separate control software for every connected device. The company has shared MHS with manufacturers and research groups working across biotechnology, robotics and quantum computing fields. Anthropic intends the preview to help establish safety checks before the standard becomes available widely to developers and equipment makers. Amazon Web Services, Automata, Danaher, Doosan Robotics, MBF Bioscience, QIAGEN, Tecan and Universal Robots are among the early participants now. Anthropic acknowledges that AI agents still require expert supervision because language models have clear limits when reasoning about physical equipment. MHS cannot currently support equipment lacking a programmable interface, leaving some laboratory and industrial machines outside its current technical scope. Anthropic says it plans to work with manufacturers to add drivers while expanding compatibility with more devices and robotics platforms. The preview will also support new safety tests and practical rules for deploying AI agents around physical equipment. The company says its eventual open-source release will include findings from these tests alongside detailed guidance for safer implementation. The project therefore remains experimental, with practical limits still depending heavily on hardware design, software access and human oversight. Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
Anthropic launches Claude Fable 5.1 and says it's up to 45 percent cheaper for agentic work
Anthropic says its newest AI models, Fable 5.1 and Mythos 5.1, address criticisms from customers about price, data retention, and overzealous safeguards. The company claims Claude Fable 5.1 offers stronger performance than Fable 5, but costs around 25 percent less typically and up to 45 percent less for complex agentic tasks, thanks to reduced pricing on cached data that was already processed and stored. Along with the announcement, a slew of early impressions popped up, including from Every CEO Dan Shipper, who claims, "It's the strongest coding model we've used, but now it's fast, token-efficient, and crucially actually speaks like a normal person." Box CEO Aaron Levie is another early access believer, saying that his company's agent with Fable 5.1 picked up on subtleties and ambiguities in data that Fable 5 missed in the same test. Meanwhile, Lisan al Gaib points out that on benchmarks, Mythos 5.1 with low reasoning scores the same as its predecessor set to Max reasoning. Fable 5.1 also has "more precise safeguards" that Anthropic says are less likely to block basic biology questions than Fable 5, but Mythos 5.1 has the same biology restrictions as the previous model. Anthropic also explained its progress on data retention, saying that Enterprise Frontier Safeguards offer "complete privacy" by storing data on the customer's cloud servers instead of its own, and will start rolling out later this fall. Anthropic says it's "now allowing Fable 5.1 to be used for identifying software vulnerabilities," but it will still redirect some cybersecurity tasks to Opus models, like "penetration testing, exploit generation, and binary-based vulnerability scanning." Claude Fable 5.1 is now available on all platforms, while Mythos 5.1 is available to Project Glasswing participants only.
Judge Rules Pentagon's Blacklisting of Anthropic Was Illegal Retaliation, Not a Security Concern
A federal court ruling this week added to the growing scrutiny around how the U.S. government is handling artificial intelligence companies that also pursue defense work. In San Francisco, U.S. District Judge Rita Lin ruled on August 27 that the Pentagon's decision to blacklist Anthropic as a national security supply-chain risk was unlawful and retaliatory rather than grounded in a demonstrated security concern. The case centers on the San Francisco-based AI company behind Claude and its dispute with the Defense Department over limits on military use of its models. Judge voids Pentagon action against Anthropic Judge Lin's 59-page order blocked the Pentagon's blacklisting of Anthropic and granted key parts of the company's summary judgment request, according to court filings and reporting from Reuters. The court found that the government's actions amounted to unlawful retaliation under the First Amendment and also denied Anthropic the due-process protections required by the Fifth Amendment. In the ruling, Lin wrote that the record did not support the government's stated justification for treating the company as a supply-chain threat. The designation had carried broad practical consequences because it effectively discouraged or blocked federal agencies and contractors from using Anthropic's technology, according to the lawsuit and subsequent court orders. Reuters reported that Anthropic sued in March after arguing the Pentagon launched an unlawful campaign of retaliation when the company refused to permit unrestricted military uses of its AI tools. The judge had already issued a preliminary injunction on March 26, signaling skepticism about the government's legal basis before issuing the final August order. Anthropic said after the ruling that it welcomed the court's decision and remains focused on working productively with the government on national security uses of AI, according to statements cited by AP and Axios. The Defense Department's legal team had argued in court that advanced AI systems are unusually difficult to evaluate and that officials needed broad discretion in assessing risk. Judge Lin rejected that rationale on the record before her, calling the challenged measures unlawful and baseless. The immediate local tie is clear: Anthropic is headquartered in San Francisco, and the ruling came from the U.S. District Court for the Northern District of California. That makes the case a significant Bay Area legal and business development, particularly for a region where AI companies, cloud providers and defense-tech firms increasingly overlap. The decision also places a San Francisco judge at the center of a national fight over how federal agencies can regulate access to emerging AI systems. What is confirmed is that the ruling removes the legal foundation for the Pentagon's supply-chain-risk designation against Anthropic. What remains unclear is whether any specific Bay Area contracts, staffing plans or federal partnerships were delayed, canceled or altered because of the blacklisting. Anthropic has not released a public accounting of any projects or agencies directly affected in California. The case still matters locally because it touches one of the Bay Area's fastest-growing sectors: generative AI companies seeking commercial growth while navigating government demand for defense applications. Washington Post and Wired both reported that the dispute grew out of a broader conflict over military access to Anthropic's systems. For local businesses and workers, the decision underscores that federal procurement fights can quickly shape the operating environment for Silicon Valley companies even when the core dispute is framed as national security policy. The larger context is the government's push to bring leading AI firms into defense and intelligence work while those same firms try to impose safety limits on how their models are used. According to Anthropic's lawsuit, the company argued it was punished after resisting Pentagon demands related to military deployment of its technology and after public scrutiny of that conflict increased. Judge Lin's earlier preliminary injunction said the measures did not appear tailored to the government's stated security interests and instead resembled retaliation. That context matters because the court did not simply disagree with a procurement choice; it found constitutional violations. Reuters reported that the order said the undisputed record showed retaliation and a lack of required pre-deprivation process. AP similarly reported that the judge concluded the supply-chain label was not supported as a genuine security measure on the evidence presented. For residents and businesses watching the case, the practical takeaway is narrow but important. Anthropic is no longer subject to the Pentagon blacklist that triggered the lawsuit, and the ruling signals that national security language alone will not automatically shield agency actions from court review when constitutional rights are at issue. Anthropic said it plans to continue working with the government on national security applications, while the broader debate over AI safeguards and military use remains unresolved in Washington and in the courts.
Sony Music, Warner Music Sue Anthropic Over AI Training
Sony Group Corp is a Japan-based company engaged in the games & network services (G&NS), music, movies, entertainment technology & services (ET&S), imaging & sensing solutions (I&SS) and other businesses. It has seven business segments. G&NS segment is involved in network service business, the manufacture and sale of home video game consoles and software. The Music segment mainly includes music production, music publishing and video media platform businesses. The Movies segment mainly includes film production, television program production and media network businesses. The ET&S field mainly includes the television business, audio, video business, still image, video camera business, smartphone business and Internet-related service business. The I&SS segment mainly includes the image sensor business. The Financial segment is involved in the insurance business and banking business. The Other segment consists of activities such as disc manufacturing business and recording media business.
Anthropic warns Claude users of infostealer malware infections that hijacked active sessions
The AI company began contacting affected users after detecting unusual usage patterns tied to stolen browser cookies across multiple malware families. Anthropic started reaching out to impacted Claude users on August 30, 2026, after discovering that infostealer malware had compromised their computers and siphoned off active login session cookies. The stolen cookies gave attackers a backdoor into user accounts, letting them burn through paid usage quotas without ever needing a password or cracking two-factor authentication. What happened and how Anthropic responded The breach came to light through unusual usage patterns on Claude accounts. Once Anthropic's security team connected the dots, they moved quickly. All compromised sessions were forcibly signed out. Saved payment methods were stripped from affected accounts. And for users who got hit with unauthorized charges from attackers running up their quotas, Anthropic issued refunds. Anthropic was also careful to draw a clear line: the malware had nothing to do with Claude itself, its infrastructure, or anything users did on the platform. The infections originated from malicious downloads and compromised software applications that users had installed on their own machines. The malware families identified in the campaign included Vidar, LummaC2, StealC, RedLine, and Acreed on Windows. A limited number of Mac devices were also affected, primarily through Atomic Stealer, also known as AMOS. Why infostealers are targeting AI accounts The infostealer families involved in this campaign are well-established tools in the cybercrime ecosystem. RedLine has been one of the most widely distributed infostealers for years, while LummaC2 has surged in popularity among threat actors for its ability to harvest browser data, crypto wallet credentials, and session tokens. Vidar and StealC operate on similar principles, scraping stored credentials and cookies from browsers. Atomic Stealer has carved out a niche as one of the more capable macOS-focused threats. The common thread across all of these tools is their focus on browser session cookies. Modern web applications keep users logged in through session tokens stored as cookies. Steal the cookie, and you inherit the session. No password prompt, no 2FA challenge, no suspicious login alert. From the server's perspective, the attacker looks identical to the legitimate user. A broader pattern of AI-related threats This session-hijacking campaign is distinct from other security concerns that surfaced around AI platforms in 2026. Earlier in the year, separate reports documented attackers misusing Claude's own features or orchestrating phishing campaigns designed to deliver malware. Anthropic's latest advisory is specifically focused on the session-hijacking aspect rather than the methods of initial infection. Some security-conscious users have started adopting browser-level protections like Google Chrome's Device Bound Session Credentials, which ties session cookies to a specific device and makes them useless if exfiltrated.
Amazon Stock Gets a Boost. AWS Adds Anthropic, Meta and OpenAI Models
This article first appeared on GuruFocus. Amazon (NASDAQ:AMZN) is widening access to artificial intelligence tools for U.S. government customers through its AWS GovCloud platform. The company said government agencies can now use several third-party AI model families through Amazon Bedrock, including offerings from Anthropic, Meta (NASDAQ:META), OpenAI, xAI and NVIDIA (NASDAQ:NVDA). Amazon's Nova models are also available within the isolated cloud environment. The expansion adds to the role of AWS in Amazon's broader AI strategy. The cloud unit reported a 37% year-over-year increase in second-quarter 2026 revenue, while operating income rose 63% from a year earlier. Amazon has pointed to AI-related workloads and its chip operations as contributors to AWS growth. Management has also said the segment could eventually generate $1 trillion in annual revenue, while noting that about 85% of worldwide information-technology spending remains on-premises. For AMZN stock, the broader AWS expansion could support longer-term growth by increasing AI adoption among government customers and other regulated users.
Amazon Stock Gets a Boost. AWS Adds Anthropic, Meta and OpenAI Models
This article first appeared on GuruFocus. Amazon (NASDAQ:AMZN) is widening access to artificial intelligence tools for U.S. government customers through its AWS GovCloud platform. The company said government agencies can now use several third-party AI model families through Amazon Bedrock, including offerings from Anthropic, Meta (NASDAQ:META), OpenAI, xAI and NVIDIA (NASDAQ:NVDA). Amazon's Nova models are also available within the isolated cloud environment. The expansion adds to the role of AWS in Amazon's broader AI strategy. The cloud unit reported a 37% year-over-year increase in second-quarter 2026 revenue, while operating income rose 63% from a year earlier. Amazon has pointed to AI-related workloads and its chip operations as contributors to AWS growth. Management has also said the segment could eventually generate $1 trillion in annual revenue, while noting that about 85% of worldwide information-technology spending remains on-premises. For AMZN stock, the broader AWS expansion could support longer-term growth by increasing AI adoption among government customers and other regulated users.
Anthropic locks out Claude users after infostealers hijack login sessions - IT Security News
Anthropic has started locking users out of their Claude accounts due to their login sessions having been compromised through infostealer malware. "The malware identified in this campaign so far include Vidar, Lumma (LummaC2), StealC, RedLine and Acreed on Windows, and Atomic Stealer (AMOS) on a small number of Macs," the company said in emails sent out to affected users last week. "It's general-purpose malware that typically arrives with an unofficial download or a malicious app, ... More →
This free, open-source browser made me ditch Perplexity Comet, Dia, and Opera Neon
Around the same time last year, AI browsers were the hot new thing. It felt like every other company wanted to rethink the browser around AI, with names like Dia, Perplexity's Comet, Opera's Neon, and even ChatGPT's Atlas all promising to change the way we use the web. While the hype has now largely died down, both from the companies building these browsers and the people actually using them, I still think there's something useful about the idea itself. Sure, the way most companies approached AI browsers often felt a little overdone and came with very genuine concerns, but having AI built directly into the browser did have its benefits. Fortunately, it turns out you don't need to hand your browsing over to one of these companies to get them. I recently came across BrowserOS, a free, open-source AI browser that promises much of what drew me to Comet, Dia, and Opera Neon in the first place, without quite as much baggage. BrowserOS is an open-source AI browser built on Chromium Chrome at heart, AI on top While there are now a bunch of AI browsers that are built on Chromium, BrowserOS is a fully open-source Chromium fork. That means the browser itself, along with the agent platform powering its AI features, is out in the open rather than being locked behind a proprietary product. This means that you have complete freedom to inspect the code, see exactly how the browser and its AI features work, and even modify or build on it yourself if you want to. BrowserOS is licensed under AGPL-3.0, and its repo even lays out the Chromium patches, build system, agent server, and browser UI that make everything work. Given that it's Chromium-based, it also feels immediately familiar. There's practically no learning curve, and importing your existing browsing data from Chrome takes seconds. All your Chrome extensions work too, and you can bring over your bookmarks, passwords, and settings in one click, which makes BrowserOS feel less like switching browsers and more like adding AI to the one you're already used to. BrowserOS doesn't lock you into one provider Pick your brain I've interacted with a lot of people who experimented with AI browsers during its initial hype, and the most frequent complaint I'd hear was privacy. Given that a browser is the one app you use to access practically everything on the internet, handing even more of that activity over to an AI company understandably made a lot of people uncomfortable. Comet keeps you within the models Perplexity chooses to support, while Opera Neon gives you a wider model selector but still largely revolves around the models and AI services Opera integrates into the browser. Similarly, ChatGPT Atlas was designed around ChatGPT itself, with OpenAI's models powering the browser's AI features. BrowserOS is much more flexible. You can bring your own API keys, use OAuth with an existing ChatGPT subscription, or skip the cloud entirely and run a local model through Ollama or LM Studio. The browser suppers over eleven providers including Anthropic Claude, Google Gemini, OpenAI, Moonshot Kimii, and OpenRouter, which gives you access to over 500 models. Now, this in no way means that BrowserOS sacrifices the actual AI-browser experience in exchange for flexibility. You still get the kind of agentic features you'd expect from the bigger names, including an AI that can understand what's on a page, navigate websites, fill out forms, extract information, and carry out multi-step tasks on your behalf. It has everything you'll find in the bigger AI browsers All the AI browser tricks are here Close First, there's the familiar AI side panel. BrowserOS lets you keep your assistant of choice open alongside whatever page you're browsing, so you can ask questions about what's on-screen, summarize something, compare information, or just use it like a normal chatbot without leaving the tab. Side-panel assistants have become almost standard across AI browsers, but BrowserOS gives you the advantage of letting you choose what's actually powering it. The more interesting part is always Agent Mode. This is where BrowserOS starts behaving much more like Comet, Opera Neon, or Atlas rather than simply sticking a chatbot next to your webpages. You can give it a task and let it actually navigate the web, click through pages, read information, and work through multiple steps on your behalf. For instance, I've used flight searches as a bit of a recurring test whenever I've tried an agentic browser. I gave BrowserOS a simple request to find flights from Karachi to Berlin for a specific set of dates. Rather than just returning a list of links or telling me where to search, it opened Google Flights in another tab, entered the route and dates itself, waited for the results to load, and then worked through the available options before summarizing the ones that made the most sense. There's no subscription attached Perplexity Comet started at $200/month. Two hundred dollars! Opera Neon spent months behind a waitlist before opening up at roughly $20 a month, while Dia introduced a $20/month Pro tier for heavier AI use. Given that BrowserOS is open-source, it's completely free to use. That said, if you choose to use it via an API key, you'll still need to shoulder whatever usage costs your chosen model provider charges. If you already pay for ChatGPT and connect through OAuth, or run a local model through Ollama or LM Studio, you can avoid adding another browser subscription on top!
Anthropic Legal Fight With Pentagon Shows Shifting Politics of AI
Welcome to Tech In Depth, our daily newsletter about the business of tech from Bloomberg's journalists around the world. Today, Austin Carr reports on Anthropic's legal case against the US government for designating the AI startup as a supply chain risk. Tech Across the Globe Meta's PR campaign: ...
China Rebukes Anthropic, Sets Terms for Key US-China AI Dialogue
China laid out conditions for artificial intelligence talks with the US, striking a strident tone ahead of a much-anticipated meeting while accusing American AI company Anthropic of bad behavior. A Sunday post on Yuyuantantian, a social media account affiliated with Chinese state broadcaster CCTV, said the US must prove its AI companies are subject to the same safety, disclosure and audit rules before any "substantive" discussions with China can take place. "A clear distinction must be drawn between genuine security threats and mere technological competition," said the post, published by an account frequently used to signal official policy thinking. "This line must be drawn jointly by all participating parties," it added. Officials from the US and China are expected to hold talks on AI before Xi Jinping's state visit to the US on Sept. 24 for a summit with Donald Trump. No date has been announced for that AI dialogue and Chinese officials last month were still seeking clarity on the agenda, in part to decide who to send, Bloomberg previously reported. Tensions are mounting around the AI race between the world's biggest economies. Underscoring that point, Yuyuantantian singled out Anthropic's Claude for criticism, accusing that model of overstepping user data boundaries, engaging in covert monitoring and transmitting website domains without authorization. "The problem is that America's own frontier models have already developed in a distorted direction," Yuyuantantian's post said. "This means the negotiation is not simply a technical dialogue from the start, but a continuation of all the earlier problems: the US is trying to turn the 'safety boundaries' it has drawn into the default rules for the entire world." The US government should be guarding security boundaries around AI, yet it has "repeatedly retreated" from doing so, according to the post titled "Anthropic Has Contracted the American Disease." Anthropic and the US State Department did not immediately respond to requests for comment sent outside US office hours. The Yuyuantantian post said the US approach to AI reflects a broader effort to contain China's technological rise. "The 'control' proposed by the US is, in essence, an attempt to make China accept an order partly defined by American companies," it added. Chinese officials are growing anxious that Anthropic's Mythos could be wielded against their country, Bloomberg previously reported. They see the potential for models like Mythos to be used as an offensive weapon, and question why Anthropic is denying China access to the systems for normal purposes, people familiar said. Anthropic blocked its services from Chinese-controlled companies last year, citing US national security concerns. But that does not seem to have effectively stopped Chinese companies from using its products. The firm, along with OpenAI and Alphabet Inc.'s Google, has expressed concern about its models being used in China. Anthropic has also accused Alibaba Group Holding Ltd. of "illicitly" accessing Claude using thousands of fraudulent accounts, an allegation the Chinese tech giant has not responded to.
Anthropic Explores How AI Could Help Improve More Capable Models
* Claude achieved 85 percent on deception tests * Its methods worked on models up to 4.7 times larger * Sonnet 5 tested more than 50 approaches in 60 hours Anthropic has published new research exploring how AI could take on a larger role in the development and safety testing of other AI systems. The work looks at whether models can independently research ways to address alignment problems, rather than relying entirely on human researchers to develop those methods. It also examines whether this approach can be used across models with different capabilities. The findings offer a closer look at how AI could eventually become more involved in AI research. Anthropic Tests Claude as an Automated AI Alignment Researcher In its research paper titled "Automated Researchers Can Reliably Mitigate Alignment Failures", Anthropic examined whether Claude could independently research ways to address 10 types of alignment failures, including deception, sycophancy, reward hacking and privacy violations. The researchers measured how much each method improved the target models while checking that their general capabilities did not decline. They also tested whether the methods worked on benchmarks Claude had not seen during its research. Claude searched existing research, proposed training methods and datasets, trained target models and evaluated the results. It could then use those results to develop another approach, allowing it to repeatedly test and refine its ideas rather than submit a single solution. Claude delivered one of its strongest results in the deception tests, closing 85 percent of the safety gap on average across several runs with Gemma-2-2B, compared with 20 percent for six experienced human safety researchers working under the same rules. Claude made more than 150 attempts during the experiment and also outperformed 28 human researchers who were given up to eight hours to develop their methods. The human comparison was not a direct test of whether Claude is a better researcher. The human participants could not repeatedly test and refine their proposals, while Claude could run multiple experiments. Anthropic instead highlighted the value of giving an AI system the ability to iterate on its own research. Some of Claude's methods also worked on alignment benchmarks that were withheld during the research process. The techniques were tested on Petri, an open-source tool that uses adversarial multi-turn scenarios to probe for misaligned behaviour. They also remained effective when applied to models up to 4.7 times larger than those Claude had originally worked on. Anthropic then tested a scenario closer to AI-assisted self-improvement. Claude Sonnet 5, which is less capable than Claude Opus 4.8, was tasked with improving the alignment of an early Opus 4.8 checkpoint. The checkpoint had not yet received most of the alignment training used for the released model. Sonnet 5 spent 60 hours testing more than 50 approaches and produced alignment results close to those of the released Opus 4.8. Its final method used just over 2,000 training examples made from simple templates and public datasets. Anthropic said the approach was roughly 15,000 times more efficient than its production alignment process. The automated system closed 65 percent of the measured safety gap in the early Opus 4.8 model, compared with 72 percent for the released version. The result does not mean Sonnet 5 improved Opus 4.8's overall capabilities or improved itself, but it shows that a weaker AI model can develop useful alignment training for a more capable system. Anthropic also tested whether its automated researchers tried to manipulate the experiments. Claude Opus 4.8 reviewed about 1,600 research-agent transcripts and flagged possible cheating in 39 cases, or 2.4 percent. Anthropic said current models often leave signs of such behaviour in their reasoning, but warned that monitoring could become harder as models become more capable. The study also has several limitations that could affect how broadly its findings apply. It covered only 10 alignment failures and did not examine issues such as political bias. Some forms of misalignment may also lack suitable benchmarks, while the researchers measured the impact on only a limited set of general capabilities. Anthropic also did not test whether the alignment gains would survive extensive reinforcement learning on other tasks. The research does not amount to full self-improving AI, since Claude did not improve or redesign itself. It instead used its research capabilities to develop alignment methods for another model, including one more capable than itself. The result offers an early indication of how AI could take on more of the research involved in developing future systems.
Anthropic Pushes Into Physical World With New Standard for AI Agents to Operate Machines
Anthropic is moving further into the physical world with a new interface designed to help AI agents communicate with and operate programmable machines. Called the Model Hardware Standard (MHS), the system is intended to work across devices used in areas such as scientific research, robotics and advanced manufacturing. Anthropic compared it to USB-C, which provides a standardized way for different devices to exchange information. The company said MHS could reduce the time and complexity involved in setting up and integrating hardware with AI systems. Unlike standards tied to a specific AI model, MHS is model agnostic, meaning organizations would not be limited to Anthropic's Claude models. The standard is currently available to a limited group of organizations through a research preview, with Anthropic planning to open-source it so manufacturers across industries can adopt the technology. The move puts Anthropic into closer competition with companies such as OpenAI and Amazon, which are also investing heavily in AI-powered hardware and physical systems. Anthropic is expanding its own hardware capabilities, including building a silicon team to develop custom chips for its AI models. Its latest initiative builds on the company's open-source Model Context Protocol, launched in 2024 to make it easier for AI agents to connect with data sources, and signals a broader push to make AI capable of interacting with the physical world.
Claude AI Users Are Getting Malware Alerts From Anthropic: Here's What You Should Know
Claude malware attack affects many users who have got an email from Anthropic telling them about its impact and what they should do. If you're a Claude user then check your email because Anthropic has a malware alert that needs to be read at the earliest. The company has informed its users that their PC might have malware that can be activated to steal personal data including passwords. More importantly, the threat is dangerous enough to allow the hacker to use their Claude account which is not a good sign. The email has been circulated across the Claude network by the company and some of them have shared the clip of the alert via Reddit. The threat is not directed at all Claude users and the company is making sure that the affected accounts are being logged out so that hackers cannot misuse their usage limits and also removing the payment method used to charge for their Claude account. You're Under Attack Anthropic's email to the affected Claude users informs them about the threat and why they may have seen a sudden surge in their usage which also results in the limits draining out faster. "If your usage limits looked like they refilled and then drained while you weren't using Claude, this was likely the cause," the email reported across channels. The malware is basically an infostealer which would have allowed the hacker to access your details on the system and even bypass any security to get Claude login sessions to get the account information for the person. Anthropic assures the leak is not linked to Claude AI by any ways, including any session you started on the AI chatbot. Phones and tablets do not appear to have been involved," the company adds, which means the PC and Mac users are the main targets for the malware attack. Be Careful What You Open Malware attacks come in all shapes and sizes but the origin of these plants are the same. You open an email or a link in a message from an unknown source, allowing the bad actor to infect the device with your consent. The security checks on the device become helpless because you have allowed the file to run and in the background the shady business has allowed your data to be exposed. Anthropic says if you are one of the victims, logging out of the account is not enough. You have to run a scan on the system for malware threats. Now you have to delink the email ID by changing the password and even setting up a two-factor authentication for better security. That's not all, the payment methods added to your Claude account should also be removed, their bank account passwords changed to make sure none of their accounts are misused.
