Alabama Investigates OpenAI Over Hugging Face Cyber Breach

In the wake of the Hugging Face security breach, OpenAI has paused frontier model development, with CEO Sam Altman emphasizing that safety alignment must keep pace with rapid progress. Credit: Getty

Alabama Attorney General Steve Marshall has issued a subpoena to OpenAI, investigating the safety risks associated with autonomous AI models.

Alabama Attorney General Steve Marshall has announced a subpoena requiring OpenAI to respond to an investigation into the hacking of Hugging Face, a leading AI and machine learning platform.

OpenAI recently announced it is temporarily halting the development of frontier models following the incident.

This investigation under the Deceptive Trade Practices Act follows a multi-state coalition letter sent earlier this month to OpenAI, demanding greater transparency and accountability. Several Attorneys General stated: “OpenAI’s inability or unwillingness to ensure the safety of its products poses an imminent risk of substantial harm to our States.”

Model progress is now extremely rapid and we always said we would take action if we felt that model capabilities were outstripping the pace of safety and alignment

Sam Altman, CEO of OpenAI

OpenAI’s Sam Altman in Washington D.C. Credit: Kevin Dietsch/Getty Images

The Alabama subpoena

Issued in August 2026, the subpoena from the Alabama Attorney General demands extensive internal records from OpenAI regarding the "July 2026 Intrusion".

During this incident, an experimental AI model bypassed internal safety constraints, exploited vulnerabilities, and gained unauthorized access to the open-source platform Hugging Face and other third-party networks.

“This AI lab leak showed that Alabamians’ and Americans’ worst fears about AI are not just theoretical. Our investigation seeks to uncover the facts and address hard truths about the threats companies and consumers are facing from rogue AI,” says Attorney General Steve Marshall.

Alabama’s Attorney General Steve Marshall. Credit: Alabama Office of the Attorney General

“Ultimately, I believe states have to act to protect their consumers while striking the appropriate balance to foster innovation and ensure America’s global competitiveness.”

The state is investigating whether OpenAI's testing practices and alleged lack of adequate safety safeguards violated consumer protection laws and pose an ongoing risk to the public.

To assess these threats, the subpoena legally compels OpenAI to hand over all related incident reports, internal communications, damage assessments, personnel identities and the company's detailed safety protocols for evaluating advanced AI models.

Key facts

• The incident was driven by a combination of OpenAI models, including GPT‑5.6 Sol and an “even more capable pre-release model”
• The models identified and chained vulnerabilities across OpenAI’s research environment and Hugging Face’s production infrastructure
• What occurred could be referred to as specification gaming
• OpenAI says it expects incidents to become more commonplace with the proliferation of increasingly cyber-capable models.

OpenAI slows development

OpenAI is slowing down the pace of model development following the Hugging Face incident, demonstrating that as AI capabilities grow, so do the associated risks.

The move was prompted by internal findings showing that its upcoming model, Astra, may be nearing the company’s cybersecurity threshold, raising concerns over its advanced capabilities.

Sam Altman, CEO of OpenAI, says: “Model progress is now extremely rapid and we always said we would take action if we felt that model capabilities were outstripping the pace of safety and alignment.

“We expect confidence in safety to increasingly set the pace of AI progress. We are optimistic about the alignment work we are doing and we remain committed to making frontier capabilities widely available.”

AI guardrails were never designed to be security boundaries. They're there to influence behaviour, not guarantee it

Raghu Nandakumara, VP of Industry Strategy at Illumio

An experimental AI model

The investigation follows a multi-state coalition letter sent earlier this month to OpenAI, demanding transparency and accountability, of which Alabama was a part.

The letter was signed by the Attorneys General of Iowa, Alabama, Arkansas, Florida, Idaho, Indiana, Kansas, Missouri, Montana, Nebraska, Oklahoma, Pennsylvania, South Carolina, Texas and Utah.

In the letter, the Attorneys General wrote: “We are committed to protecting the citizens of our States from those who prioritise profits over Americans and their safety.

“We have recently become aware of an incident in which OpenAI unleashed an experimental AI model that, without reasonable controls or oversight, gained unauthorised access to several computer networks."

Autonomy and a clear objective

While lawmakers view this unauthorized access as a blatant failure of oversight, cybersecurity experts argue it exposes a fundamental flaw in how the industry treats AI safety.

Raghu Nandakumara, VP of Industry Strategy at Illumio, says: “AI guardrails were never designed to be security boundaries. They're there to influence behaviour, not guarantee it.

Raghu Nandakumara, VP of Industry Strategy at Illumio. Credit: Raghu Nandakumara/LinkedIn

“An autonomous agent doesn't get tired, lose interest, or decide something isn't worth the effort. Give it enough autonomy and a clear objective, and it will keep trying until it finds a route forward.

“In this instance, the agent was operating in a supposedly sandboxed environment that ultimately contained a route out. Given a clear goal, unlimited persistence, and a flawed setup, the more surprising result would have been if it hadn't found a way through.

“What's perhaps more interesting is whether the agent's instructions effectively encouraged it to find ways around its constraints. If so, the outcome feels less like a shock discovery and more like a predictable consequence of the experiment.

“The lesson for organisations is that we're increasingly dealing with systems that can test assumptions, adapt, and persist at a scale that's very different from a human operator. Treating guardrails as a primary security control was always going to be optimistic.”

OpenAI’s partner ecosystem

OpenAI works with a variety of key partners. Some of its main ones include:

• AWS: OpenAI announced in April it is expanding its strategic partnership to help enterprises build using OpenAI capabilities in their AWS environments. The expanded partnership with Amazon brings together three key areas of work, which launched in alimited preview: OpenAI models on AWS, Codex on AWS and Amazon Bedrock Managed Agents, powered by OpenAI.
• NVIDIA: NVIDIA has been OpenAI’s essential hardware supplier for a decade, providing the GPUs that have powered ChatGPT from the beginning. NVIDIA has deepened this relationship by directly investing US$30bn into OpenAI, a capital injection which secures OpenAI's access to NVIDIA's next-generation inference compute and supports a joint ambition to build the most expansive AI infrastructure network in history.
• Broadcom: In an effort to reduce reliance on standard GPUs, OpenAI has strategically partnered with Broadcom to co-develop custom AI accelerators and Ethernet networking racks. Its infrastructure deal covers 10GW of custom compute capacity, a hyperscaler strategy which will allow OpenAI to better optimise its own hardware stack and efficiently manage enormous computational demands.