Anthropic plans to expand access to its Mythos model, but the US government opposes it.

White House tensions with the AI company Anthropic have escalated once again. Amid ongoing cybersecurity concerns, **the Trump administration has explicitly opposed expanding the scope of Anthropic’s AI model Mythos**, a stance that not only blocks the commercial rollout of the model but also reflects the deep structural conflicts between the government and Silicon Valley AI companies. According to The Wall Street Journal, citing sources familiar with the matter, Anthropic recently proposed to grant about 70 additional enterprises and institutions access to Mythos, which would bring the total number of entities permitted to use the model to around 120. Trump administration officials have voiced clear opposition, citing security concerns. Some White House officials also worry that Anthropic may lack adequate computing power to simultaneously serve such a large number of users, potentially affecting the government’s effective use of the model. **This confrontation occurs against the backdrop of already heightened tensions between the two sides.** Anthropic and the White House previously had public disputes over Pentagon AI tool usage, with related lawsuits proceeding in two different courts. Meanwhile, Anthropic announced last week that it is investigating a suspected unauthorized access incident involving Mythos, further fueling concern over the spread of large-scale software vulnerabilities and amplifying market attention to AI security risks. **Mythos Security Concerns: Vulnerability Exploitation Raises Alarm** The core reason for the White House’s intense focus on Mythos lies in its outstanding capabilities in cyber offense and defense. Cybersecurity experts point out that **the model can detect and exploit software vulnerabilities, possesses the potential to launch cyberattacks, and can cause large-scale damage on networks—abilities that have deeply unsettled multiple government agencies and businesses.** In early April this year, Anthropic made a preview version of Mythos available to select top enterprises and institutions, immediately sparking global concerns and prompting many to begin evaluating and guarding against the potential safety risks of broader model deployment in the future. Currently, initial access to the model covers about 50 organizations responsible for managing key infrastructure. The Trump administration is working to expand government agency access, but does not support broader commercial rollout. Leading AI firms, including OpenAI and Google (under Alphabet), have also provided early access to their models for security researchers so vulnerabilities can be discovered and patched before they are exploited maliciously. However, according to The Wall Street Journal, industry insiders are troubled by the sheer number of vulnerabilities found, as well as the workload involved in creating, testing, and installing patches. **Deep rifts: Political Differences and Personnel Disputes** The conflicts between the White House and Anthropic are not limited to security; **political disagreements are also significant friction points**. The Trump administration has criticized Anthropic, partly due to its ties with liberal funders who support strict AI regulation, and partly because it employs several former Biden administration officials. **Personnel issues are similarly specific and obvious.** According to reports citing knowledgeable sources, former Anthropic researcher Collin Burns was initially favored for a government position dedicated to evaluating top AI models, but the government changed its plans and chose someone else. Sources say that senior White House officials were not informed in advance about Burns’s potential appointment and were reluctant to see individuals from leading AI companies take roles requiring close collaboration with those companies. **Computing Power Bottleneck: Practical Constraints on Commercial Expansion** Aside from political dynamics, **limited computing power is also a genuine obstacle for Mythos expansion**. Some White House AI advisors, including venture capitalist David Sacks, have speculated that Anthropic’s selective rollout is partly due to its comparatively limited computing resources, which are critical for running various tasks for AI model users. To address this shortfall, Anthropic has recently reached agreements with Amazon, Google, and Broadcom to increase computing power supply, but these projects need time to become fully operational. AI companies typically prioritize government and other specific user groups when allocating computing resources, while the industry hopes for more efficient models to fundamentally ease the computing crunch. **Calls for Cooperation: Real-world Pressures Amid Differences** Despite ongoing conflicts, both sides are calling for cooperation. Dean Ball, a former Trump administration AI advisor and now senior fellow at the American Innovation Foundation, noted that **the opposition between Anthropic and the Trump administration is counterproductive. Cooperation is necessary on core issues such as cybersecurity, and confrontation will only backfire.** White House officials also state that the government is balancing innovation and security, and actively collaborating with the private sector to ensure safe deployment of AI models. This statement leaves some room for relationship repair, but for now, the White House’s decision to refuse Anthropic’s expansion of Mythos access clearly shows ongoing tension and unresolved differences. Risk notice and disclaimer: The market involves risks; investment needs caution. This article does not constitute personal investment advice and does not take into account individual users’ specific investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their individual circumstances. Invest accordingly, at your own risk.