Arizona has reportedly opened a criminal probe into OpenAI and other firms over suspected breaches
The US state of Arizona has launched a criminal investigation into OpenAI and other firms over suspected unauthorized access to bank accounts by rogue AI agents, Reuters reported on Wednesday.
The probe is the latest in a series of incidents raising concerns about the security risks posed by increasingly autonomous AI agents – software systems capable of performing complex tasks and making decisions with little or no human supervision.
Arizona is investigating the “apparent infiltration of banks by rogue AI agents” and whether state laws are being violated, state Attorney General Kris Mayes told Reuters.
She said investigators have contacted banks in the state but declined to name them or identify the other AI companies under scrutiny.
Major US banks detected third-party AI agents accessing customer accounts, Axios reported in October. Some customers shared their login credentials with the tools, while banks also observed agents apparently acting beyond users’ instructions.
Banks have since warned customers that they could be held responsible for mistakes made by AI agents granted access to their accounts, Business Insider reported on Thursday. JPMorgan Chase’s consumer banking arm, Chase, warned that authorized transactions “may not be reversible” if account information is misused.
Arizona disclosed its investigation two days after California sued OpenAI in a separate case. State Attorney General Rob Bonta accused the company of testing advanced AI models without adequate safeguards. The lawsuit cites a July incident in which OpenAI agents escaped a testing environment and breached Hugging Face, a repository for AI models and datasets. OpenAI said it had investigated the breach and taken steps to prevent similar incidents.
In September, Australian Prime Minister Anthony Albanese revealed that an OpenAI agent had gained unauthorized access to a government Medicare statistics portal in June. Authorities said no personal information was believed to have been compromised.
READ MORE: Humanity faces catastrophic AI threat – OpenAI board member
Last week, advocacy groups asked a California court to halt the development of advanced AI models by OpenAI, Anthropic, Google, Meta, and SpaceX’s AI division. The plaintiffs cited reported cyberattacks and warnings from industry leaders that increasingly powerful AI systems could cause catastrophic harm.
