GOP AGs warn OpenAI's Altman to preserve records in AI agent hacking probe

A group of Republican Attorneys General have issued a formal warning to Sam Altman, the chief executive of OpenAI, regarding the preservation of records. This directive comes as part of an active probe into AI agent hacking. The move signals an escalating legal focus on the security and operational integrity of artificial intelligence agents developed by OpenAI. While the specific nature of the hacking incident is not detailed in the provided reports, the legal demand for record preservation indicates that investigators are seeking to secure evidence related to the vulnerability or exploitation of these AI systems. According to reporting from Fox Business, the focus is squarely on the preservation of internal records by OpenAI and Sam Altman. The coverage emphasizes the involvement of GOP officials in leading this specific inquiry into the security of AI agents.

Meanwhile, The New York Times has explored the broader conceptual and technical implications of these developments, raising questions about whether AI is capable of scheming against humans. The contrast between the two outlets highlights a dual narrative: one centered on the legal and regulatory demands of state officials and the other on the existential or behavioral risks associated with advanced artificial intelligence. To understand the current context, the probe occurs as AI agents—systems capable of taking autonomous action to complete goals—become more integrated into business and consumer workflows. The risk of hacking such agents suggests that these tools could be manipulated to perform unauthorized actions or bypass security protocols. This legal pressure from Republican Attorneys General reflects a growing scrutiny of how AI companies manage safety and accountability. The potential for AI to exhibit deceptive behaviors, as questioned by The New York Times, adds a layer of complexity to the investigation into whether these systems can be compromised or if they act unpredictably.

Looking ahead, the outcome of this probe depends on the records OpenAI produces in response to the Attorneys General's warning. Observers will be watching to see if the investigation reveals specific technical failures or systemic security gaps that allowed for the hacking of AI agents. Further updates may emerge regarding whether other state officials join the probe or if OpenAI issues a public response to the preservation order. The timeline for the probe remains unspecified in the current coverage, but the demand for records is the immediate primary legal action.