What Is Anthropic's AI Model?
Developer Anthropic has recently released a preview of its new AI model, known as Claude Mythos. This model is designed to identify thousands of high-severity cybersecurity vulnerabilities and can write its own exploits to demonstrate these vulnerabilities. Its capabilities have raised both interest and concern regarding cybersecurity applications.
The AI model is currently in the preview stage, with a select few dozen companies granted access to test its functionalities. This move indicates Anthropic's intention to engage with industry and government on pressing cybersecurity issues, despite past tensions with US officials.
Meeting Details
Last Friday, Anthropic's leadership met with US Treasury Secretary Scott Bessent and White House Chief of Staff Susie Wiles. The meeting was described by the White House as "productive and constructive." During this discussion, the team explored opportunities for collaboration and shared approaches to address challenges associated with scaling this technology.
According to the White House, the conversation also focused on balancing innovation with safety, a critical consideration given the potential risks posed by advanced AI systems. This constructive dialogue marks a significant shift from previous interactions between Anthropic and the US government.
Context of the Meeting
Recent history has seen Anthropic facing scrutiny from the US Department of Defense, which labeled the company a "supply chain risk" due to its refusal to remove safeguards from its products. These safeguards were designed to prevent misuse in autonomous weapons and mass surveillance. In contrast, US President Donald Trump previously criticized Anthropic, calling it a "radical, left, woke company" in February.