The White House is reportedly preparing to implement a new AI model, Mythos, developed by Anthropic, across major federal agencies. A memo from Gregory Barbaccia, the federal chief information officer of the White House Office of Management and Budget, indicates that this distribution is in progress. Key departments such as Defense, Treasury, Commerce, Homeland Security, Justice, and State have been informed about the initiative.

According to the memo reviewed by Bloomberg, the Office of Management and Budget is in the process of establishing safeguards to enable the use of Mythos by these agencies within the coming weeks. However, it is important to note that these plans are still subject to change.

Previously, a leak suggested that Anthropic was developing a powerful and potentially dangerous AI model named Mythos, a claim that the company later confirmed. Anthropic raised alarms about the model's capabilities, warning that it could pose cybersecurity threats if misused. As a precaution, the company decided to delay the public release of Mythos and instead introduced Claude Opus 4.7, which it described as less advanced than Mythos, particularly regarding its cyber functionalities.

While the general public received Claude Opus 4.7, select tech giants, including Nvidia, Microsoft, Google, and Apple, as well as financial institutions like JPMorgan Chase, were granted limited access to a preview version of Mythos through an initiative called Project Glasswing.

Reports have emerged detailing the concerns among government and corporate officials who had the opportunity to preview Mythos. For instance, the Bank of England has initiated urgent discussions with cybersecurity experts following their review of the model, as reported by the Financial Times. Moreover, the Information highlighted that the cryptocurrency sector is actively seeking access to Mythos to bolster defenses against potential threats to cryptography posed by the model.

Bloomberg’s report indicates that Anthropic had briefed senior officials in the U.S. government about Mythos before granting limited access to corporate partners and other countries. These briefings reportedly covered the model's potential applications for both offensive and defensive cyber strategies.

Interestingly, the Trump administration's willingness to integrate Anthropic's Mythos into federal operations marks a significant shift, given that just over a month ago, the company's products were classified as national security risks and effectively banned from federal usage. This unprecedented decision was historically reserved for foreign companies and followed a dispute between the Pentagon and Anthropic regarding the military applications of its models, coinciding with the timing of U.S. military actions in Iran.

Additionally, shortly after Bloomberg's report, Politico revealed that the Commerce Department’s Center for AI Standards and Innovation had begun evaluating Mythos's capabilities even before Anthropic officially acknowledged its existence. The report further claims that personnel from at least two additional large federal agencies had proactively reached out to Anthropic for access to Mythos, despite the existing ban, and that at least three congressional committees had sought briefings from the company in the past week.

Source: Gizmodo News