Anthropic eyes $900B+ valuation in accelerated fundraise

The White House is unlikely to extend export restrictions on Anthropic's advanced AI models to other AI companies, according to a government official. The restrictions on Anthropic stemmed from the company's refusal to address vulnerabilities in its Fable 5 and Mythos models. This signals a narrow approach to export controls that targets specific companies rather than broad sectoral restrictions.

The U.S. government ordered Anthropic on Friday to block all access to its Fable 5 and Mythos 5 models for foreign users and employees, citing national security concerns. Anthropic complied by cutting off access entirely for all customers. The company stated the government did not provide specific details about the security threat, only verbal evidence of potential jailbreak vulnerabilities that Anthropic characterizes as minor and duplicative of issues in other models.

Anthropic is pursuing its first data center leases to control its own server infrastructure for AI development, aiming to reduce computing costs long-term. The company has signed more than a dozen letters of intent with U.S. data center developers. Anthropic is also in discussions with Google about a financial guarantee arrangement for lease payments, with Google potentially co-designing server chips for the facilities.

Anthropic CEO Dario Amodei published a policy essay calling for FAA-style government regulation of powerful AI models, arguing that mandatory third-party testing and potential deployment blocks are necessary as AI capabilities grow. The company released two policy frameworks alongside its most powerful model yet, Claude Fable 5, and a gated version with advanced cyber capabilities. The proposal would require models trained with over 10^25 FLOPs or developed by companies with $500 million in AI revenue to undergo mandatory testing, with regulators able to block or delay deployment if severe risks are identified.