The US government has banned Anthropic's latest model, and state attorneys general are jointly investigating OpenAI.

(San Francisco) The U.S. artificial intelligence (AI) industry is facing an unprecedented regulatory storm. Two leading AI companies, Anthropic and OpenAI, were hit hard by the federal government and a coalition of state attorneys general on Friday (June 12).

On the one hand, the U.S. government issued an emergency export control order on national security grounds, forcing Anthropic to shut down its most advanced large language model globally; on the other hand, a coalition of attorneys general from multiple states launched a joint investigation into OpenAI and issued civil subpoenas for alleged misuse of consumer data and endangerment of minors.

Anthropic issued a statement on Friday saying that at 5:21 p.m. that day, the company received instructions from the government to prohibit foreigners, including the company's foreign employees, from using its two latest AI models, Fable 5 and Mythos 5. The two models were just released on June 9.

Anthropic said the government did not specify its national security concerns, but the company's understanding is that "the government believes there is a way to bypass or 'crack' the Fable 5 model, and that this method could be used by hackers to launch attacks."

Anthropic has complied with instructions to stop making these two AI models available to all customers, but expressed strong reservations about this practice. The statement pointed out that removing a commercial model used by hundreds of millions of people simply because of a potential minor flaw "if this standard were applied to the entire industry, we believe that almost all leading model vendors would be unable to release new models."

This latest move marks a significant escalation in the U.S. effort to curb foreign adversaries' access to advanced AI capabilities. In recent years, U.S. export controls have primarily targeted high-end chips and development tools that power AI, rather than directly restricting foreign use of AI models.

According to a report by The Wall Street Journal, sources familiar with the matter said that attorneys general from multiple states have joined forces to launch a broad investigation into OpenAI, and on Friday issued subpoenas to OpenAI, demanding that it submit a large number of documents related to its operations and user impact. The subpoenas reportedly cover advertising operations, user activity and retention, consumer and health data management, products and services targeting minors and the elderly, deep learning models, the issue of "model sycophancy," and internal company rules and regulations.

OpenAI issued a statement saying, "AI is an emerging and powerful technology, and we are committed every day to ensuring that the public can safely and responsibly enjoy its benefits. We value the concerns raised by state attorneys general and plan to engage in constructive communication with their offices."

Earlier this month, Florida became the first state to sue OpenAI and its CEO, Altman, accusing them of knowingly releasing the chatbot ChatGPT despite its safety risks and ignoring warnings of potential harm to users. The lawsuit alleges that ChatGPT was used to assist in planning school shootings, encourage self-harm, and cause addiction among minors.

Source: [Lianhe Zaobao] (https://www.zaobao.com/news/world/story20260613-9201812)