Regulatory risks are increasing for U.S. artificial intelligence (AI) companies OpenAI and Anthropic as they prepare for initial public offerings (IPOs). OpenAI is under investigation by U.S. regulators, while Anthropic faces restrictions on the overseas use of its advanced AI models.
According to reports from the Wall Street Journal and other outlets, OpenAI received a subpoena from the New York Attorney General's office on June 12 related to an ongoing investigation by a coalition of state attorneys general. This marks the latest legal action against an AI company by state authorities.
The subpoena reportedly covers various aspects of OpenAI's operations, including its advertising practices, user engagement and retention strategies, handling of consumer and health information, and policies for protecting minors and the elderly. It also requests documentation related to the development of AI models and internal policies, including issues surrounding model alignment, which refers to the phenomenon where AI excessively aligns with user opinions or preferences.
Regulatory pressure is also mounting on Anthropic. On the same day, the U.S. government announced export control guidelines for Anthropic's latest AI models, 'Fable5' and 'Mythos5.' Authorities believe that some researchers could bypass safety measures in Fable5 through specific prompts, leading to its inclusion in the export control list.
In response, Anthropic has restricted user access to these models to comply with the new regulations. The company stated, "Claims that Fable5 can be circumvented are a misunderstanding," and added that it is working to restore services.
These developments illustrate the increasing scrutiny the U.S. government is placing on AI technologies. As national security and safety concerns become central issues in the AI industry, the scope of government regulation is expanding from the models themselves to the overall operation of services.
This trend aligns with recent actions by the U.S. government to tighten controls on AI technologies. On June 2, President Donald Trump signed an executive order allowing the government to conduct security reviews for up to 30 days before the release of new AI models. The Software Policy Research Institute (SPRi) noted that this move stems from concerns that rapid advancements in AI performance could pose national security threats.
Meanwhile, Sam Altman, CEO of OpenAI, has postponed his visit to South Korea. Altman was originally scheduled to attend the Samsung 'Device Experience (DX) Insight Talk #2' event on June 14-15, where he was to discuss AI collaboration with Kakao CEO Jeong Jin-ho and visit Naver's second headquarters, '1784.' However, all related plans have been delayed due to the postponement of his trip.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.