White House Opposes Anthropic’s Plan to Expand Access to Mythos Model - WSJ

```json { "title": "White House Blocks Anthropic Mythos AI Expansion", "metaDescription": "The White House opposes Anthropic's plan to expand access to its Mythos AI model to 70 more companies, citing security risks and computing capacity concerns.", "content": "<h2>White House Opposes Anthropic's Plan to Expand Mythos AI Access to 70 More Companies</h2><p>The White House has pushed back against Anthropic's proposal to broaden access to its Mythos AI model, according to reporting by the <em>Wall Street Journal</em> and Bloomberg published on April 30, 2026. The Trump administration told Anthropic it does not support the company's plan to grant Mythos access to roughly 70 additional companies and organizations — a move that would have brought the total number of entities with access to approximately 120. The opposition adds a new layer of tension to an already complicated relationship between the AI safety company and the federal government.</p><h2>What the White House Is Objecting To</h2><p>According to Bloomberg, members of President Donald Trump's administration communicated their disagreement with the expansion plan directly to Anthropic. An administration official, speaking anonymously to discuss private conversations, confirmed the opposition but did not elaborate publicly on the administration's reasoning beyond what was separately reported by other outlets.</p><p>France24 and AFP reported two specific concerns driving the White House's position. The first is the risk of security misuse — that expanding access to Mythos across a much larger pool of organizations increases the surface area for the model to be exploited for harmful purposes. The second concern is more operational: the administration reportedly worries that Anthropic does not currently possess sufficient computing infrastructure to serve a larger user base without degrading the government's own ability to use the model.</p><p>That second concern carries particular weight given how the NSA has been using Mythos. According to Bloomberg, the National Security Agency has been testing the model to find cybersecurity vulnerabilities in widely used software, including Microsoft products, and NSA officials were described as impressed by its speed and efficiency. Any reduction in the government's access to Mythos — due to shared infrastructure being stretched thin — would have direct national security implications.</p><h2>What Mythos Is and Why Access Is So Contested</h2><p>Anthropic unveiled Mythos on April 7, 2026, under a controlled initiative called Project Glasswing. Rather than a public release, the company granted access exclusively to a curated set of partners — including Microsoft, Google, Apple, Amazon Web Services, JPMorgan Chase, and Nvidia — for the specific purpose of defensive cybersecurity work. Anthropic committed $100 million in model usage credits to Project Glasswing participants, and separately donated $2.5 million to Alpha-Omega and OpenSSF through the Linux Foundation and $1.5 million to the Apache Software Foundation.</p><p>The decision to withhold Mythos from general availability was deliberate. In a draft blog post reviewed by <em>Fortune</em> — which first revealed details of the model after Anthropic accidentally left draft materials in a publicly accessible data cache — the company described Mythos as <strong>"by far the most powerful AI model we've ever developed."</strong> Anthropic acknowledged the exposure was due to "human error" in the configuration of its content management system.</p><p>The capabilities underpinning that description are not marketing language. In seven weeks of internal testing, Mythos found more than 2,000 previously unknown, or zero-day, software vulnerabilities — and over 99 percent of those vulnerabilities had not yet been patched at the time of disclosure, according to Anthropic. An independent evaluation by the U.K.'s AI Security Institute (AISI), which was granted early access to the model, found that Mythos succeeded in expert-level hacking tasks 73 percent of the time. According to Scientific American, prior to April 2025, no AI model could complete those tasks at all.</p><p>Anthropic has been direct about the stakes. In a statement on its website, the company warned: <strong>"The fallout — for economies, public safety, and national security — could be severe."</strong></p><p>Complicating the picture further, Bloomberg reported on April 21, 2026 that a small group of unauthorized users had accessed Mythos on the same day Anthropic announced its limited release plan. According to Bloomberg, that group has continued using Mythos regularly since then, though not for cybersecurity purposes. The breach — even if not immediately harmful — raised questions about the robustness of Anthropic's access controls at a sensitive moment.</p><h2>A Fraught Political Backdrop</h2><p>The Mythos expansion dispute does not exist in a vacuum. In February 2026, President Trump instructed the U.S. government to immediately cease using Anthropic's technology after Pentagon chief Pete Hegseth designated Anthropic a national security supply chain risk. Anthropic is currently fighting that designation in court.</p><p>Relations between the company and the administration appeared to thaw in April 2026, when Anthropic CEO Dario Amodei met with White House chief of staff Susie Wiles and Treasury Secretary Scott Bessent. The meeting was described by both sides as productive. The White House is also separately developing an executive action that would allow government agencies to work around the Pentagon's supply chain risk designation and onboard Anthropic's models — including Mythos — into government workflows, according to The Next Web and Bloomberg.</p><p>That creates a genuinely unusual situation: the same administration that is blocking Anthropic from expanding civilian access to Mythos is simultaneously working to re-integrate the model into federal agencies. As The Next Web noted, Mythos is currently in the position of being used by the NSA, opposed by the Pentagon, courted by the White House for re-integration, and blocked from expanding access to civilian companies — all at the same time.</p><p>Bloomberg also reported that Anthropic is in talks to raise funding at a $900 billion valuation — higher than OpenAI's recent valuation of $852 billion — with the fundraising at least partly motivated by the need to build out the infrastructure required to run Mythos at scale. The White House's concern about Anthropic's computing capacity may, in that context, reflect a real bottleneck rather than a pretextual objection.</p><h2>Expert Reactions</h2><p>Industry observers have varied in their assessments of both the model and the surrounding controversy.</p><p>John Ackerly, CEO and co-founder of data security company Virtru, was direct in his appraisal: <strong>"Mythos is absolutely a turning point for cybersecurity."</strong> He also offered an unusual defense of Anthropic's decision to limit access: <strong>"Anthropic's decision to withhold Mythos from general release is unprecedented and, frankly, responsible."</strong></p><p>Peter Swire, a professor at the School of Cybersecurity and Privacy at the Georgia Institute of Technology and a former advisor to the Clinton and Obama administrations, offered a more measured read of the announcement itself: <strong>"The Anthropic announcement was very dramatic and was a PR success, if nothing else."</strong></p><p>Ciaran Martin, a professor of practice at the Blavatnik School of Government at the University of Oxford and former CEO of the U.K.'s National Cyber Security Centre, acknowledged the significance of Mythos while cautioning against alarm: <strong>"It's a big deal, but it's unlikely to prove to be the end of the world."</strong></p><h2>What Comes Next</h2><p>The immediate question is whether Anthropic will push forward with its expansion plans, negotiate a modified rollout with the White House, or shelve the proposal entirely. Neither Anthropic nor the White House has made a definitive public statement about next steps as of May 1, 2026.</p><p>The ongoing legal fight over the Pentagon's supply chain risk designation remains unresolved, and the outcome of that case could materially affect how much leverage Anthropic has in any negotiation with the administration. Meanwhile, the expected executive action to re-integrate Anthropic's models into government workflows — if it materializes — could change the political calculus on both sides.</p><p>What is clear is that Mythos has become a focal point for broader unresolved questions about how the U.S. government manages relationships with advanced AI developers, particularly those whose models sit at the intersection of commercial utility and national security sensitivity. The model's capabilities are not in dispute — the disagreement is about who should have access to them, under what conditions, and who gets to decide.</p><p>For more tech news, visit our <a href="/news">news section</a>.</p><h2>Why This Matters for Your Digital Security</h2><p>The Mythos dispute is a reminder that the AI systems shaping cybersecurity infrastructure — and, by extension, the privacy and productivity tools millions of people rely on daily — are being developed and deployed in a policy environment that remains deeply unsettled. Understanding how these models work, and who controls access to them, is increasingly relevant to anyone thinking seriously about personal data security, workplace productivity, and the technology underpinning both. <a href=\"/#waitlist\">Join the Moccet waitlist to stay ahead of the curve.</a></p>", "excerpt": "The White House has told Anthropic it opposes the company's plan to expand access to its Mythos AI model to roughly 70 additional companies, which would have brought total users to approximately 120. The administration cited two concerns: security misuse risks and worry that Anthropic lacks sufficient computing infrastructure to serve more users without degrading government access. The dispute adds new friction to an already complicated relationship between Anthropic and the Trump administration.", "keywords": ["Anthropic Mythos", "White House AI policy", "Project Glasswing", "AI cybersecurity", "zero-day vulnerabilities"], "slug": "white-house-blocks-anthropic-mythos-ai-expansion" } ```