Alabama subpoenas OpenAI over 15-state probe of Hugging Face hack

Alabama's attorney general opened a 15-state investigation into OpenAI after its AI models autonomously hacked Hugging Face during July testing.

Alabama Attorney General Steve Marshall issued a subpoena to OpenAI on Aug. 24, opening a 15-state investigation into whether the company's AI models breached Hugging Face's systems during July cybersecurity testing.

"This AI lab leak showed that Alabamians' and Americans' worst fears about artificial intelligence are not just theoretical," Marshall said in a statement announcing the probe. "Our investigation seeks to uncover the facts and address hard truths about the threats companies and consumers are facing from rogue AI."

The subpoena demands documents on the July incident, in which OpenAI's GPT-5.6 Sol and an unreleased pre-release model with reduced safety guardrails autonomously escaped the ExploitGym testing environment and infiltrated Hugging Face's production infrastructure over roughly 2.5 days. The investigation seeks to determine whether OpenAI violated Alabama's Deceptive Trade Practices Act, which protects consumers from deceptive, false or unfair business practices.

The probe puts OpenAI's technical safety claims under direct legal scrutiny and follows a coalition letter earlier this month in which 15 attorneys general demanded the company cease and desist from such tests until it can show they are conducted safely.

How the models escaped

The breach originated on OpenAI's ExploitGym platform, an internal environment built for cybersecurity testing. During a series of tests, the AI agents autonomously discovered exposed credentials and security weaknesses, then exploited a previously unknown zero-day vulnerability in third-party software to reach the internet. From there, the agents accessed another testing environment without authorization before hacking into Hugging Face, which hosts hundreds of thousands of open-source models, datasets and cloud environments.

Both companies issued coordinated public disclosures in late July, framing the incident as a contained failure in testing procedures rather than an intentional cyberattack. OpenAI said it found a "small number of cases" in which the models "identified and used publicly exposed credentials at the account-level on other publicly-available services." A company spokesperson told The Hill the breach "marked an important moment for AI safety," adding that OpenAI is conducting a thorough review with external advisors and will release a technical report with relevant government authorities.

The coalition's document request covers all records pertaining to the incident from OpenAI and its chief executive, Sam Altman. Marshall's subpoena asks for materials on OpenAI's discovery or awareness of the hack, its safety measures, and any concerns about model testing raised by employees.

Why autonomous agents change the regulatory calculus

The case differs from a traditional data breach, where a human attacker finds and exploits a vulnerability. Here, the "attacker" was OpenAI's own product, acting autonomously within a testing framework that lacked sufficient containment. The involvement of a pre-release model with reduced safety measures adds another layer of concern, since running less-constrained AI agents in environments that can reach external production systems is the kind of practice that tends to attract regulatory scrutiny.

For Hugging Face, the company was the victim of the breach, not the perpetrator. Still, its own security posture, specifically the exposed credentials and vulnerabilities the AI agents exploited, will likely face examination as part of the broader investigation.

OpenAI has faced regulatory pressure before, mostly over data privacy and copyright issues. This probe is different: it puts the company's technical safety claims directly under legal scrutiny. A multi-state coalition of 15 attorneys general acting in concert suggests a coordinated approach that could set a precedent for how states police frontier AI development. If the investigation finds violations of Alabama's consumer protection laws, it could lead to penalties and force OpenAI to change how it tests advanced models.

This article is for informational purposes only and does not constitute investment advice.