On Friday, the U.S. administration signed a directive that effectively paralyzed two of Anthropic's flagship models — Fable 5 and Mythos 5. The reason is a ban on foreign users accessing these AI systems. The company was forced to instantly disable the models for non-American clients, triggering a sharp reaction from Anthropic's leadership.
By Saturday, the company's senior engineers held emergency online talks with White House representatives. According to my data, Commerce Secretary Howard Lutnick and Cybersecurity Director Sean Kerncross participated in the dialogue. Anthropic was represented by co-founder and head of computing Tom Brown, as well as policy director Sarah Hack.
The main goal of the talks is to secure the lifting of export restrictions, which have essentially deprived international users of access to the company's most advanced developments. The situation is compounded by the fact that both sides are interested in resuming model operations as soon as possible, but a specific mechanism for unblocking has yet to be agreed upon.
One option under consideration involves Anthropic engineers joining forces with security personnel to find a compromise. However, this is only a temporary measure. The regulatory pressure on Anthropic is not an isolated incident. It is part of a broader trend where governments attempt to control access to cutting-edge AI technologies, fearing their use for hostile purposes.
Notably, Anthropic recently confidentially filed an S-1 with the SEC, signaling preparations for a public offering. A prolonged conflict with authorities could negatively impact investor perception of the company and delay its IPO. The market is closely watching developments: if restrictions are not lifted soon, it could lead to a reassessment of Anthropic's valuation.
Expert commentary: The situation surrounding Fable 5 and Mythos 5 is a classic example of how regulatory uncertainty stifles innovation. Anthropic finds itself in a trap: on one hand, it must comply with directives; on the other, it risks losing the international market and investor trust. If no compromise is reached, we may see either a relaxation of the rules or, conversely, tighter control over the entire AI industry.