Anthropic informs Trump: tensions surrounding the Mythos model

Anthropic and the Trump Administration: Tensions Surrounding the AI Model Mythos

Jack Clark, co-founder of Anthropic and head of public benefits for Anthropic PBC, recently confirmed that the company informed the Trump administration about its artificial intelligence model, Mythos. This model, announced last week, is considered too dangerous to be made available to the general public, primarily due to its advanced cybersecurity capabilities.

In an interview at the Semafor Global Economy Summit, Clark explained why Anthropic continues to collaborate with the U.S. government while also being in litigation with it. In March, Anthropic filed a lawsuit against the Department of Defense of the Trump administration after the agency labeled the company a risk to the supply chain. The dispute centered on the unlimited access the Pentagon wanted to have to Anthropic's AI systems for uses such as mass surveillance and autonomous weapons. Ultimately, OpenAI won the contract instead of Anthropic.

Clark downplayed the significance of this risk designation, describing it as a "narrow contractual dispute" and asserting that it should not overshadow the company's commitment to national security. "Our position is that the government needs to be informed about these issues, and we need to find new ways for the government to partner with a private sector that is creating things that truly revolutionize the economy, but that will have aspects touching on national security, equities, and other areas," he stated. Clark also confirmed that discussions had taken place with the Trump administration regarding Mythos and other future models.

This confirmation comes after reports indicated that officials from the Trump administration were encouraging banks such as JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley to test Mythos.

Clark also addressed the broader impact of AI on society, particularly concerning unemployment and higher education. While Anthropic's CEO, Dario Amodei, has expressed concerns about potential mass unemployment similar to the Great Depression, Clark nuanced these remarks. He explained that Amodei believes AI will progress faster than expected, which underpins his estimates.

Leading a team of economists at Anthropic, Clark noted that the company has so far observed only "a few potential weaknesses in the employment of recent graduates" in certain industries. He emphasized that Anthropic is preparing for possible major changes in the labor market.

When asked about the academic paths students should consider or avoid in light of the impacts of AI, Clark suggested that important fields are those that "involve synthesis across a variety of subjects and analytical thinking." "This is because what AI allows us to do is to have access to a sort of arbitrary amount of subject matter experts in different fields," he added, highlighting the importance of asking the right questions and having insights into the intersections of ideas across disciplines.