Updated ETGet the DI Daily by Email

Technology news for e-commerce professionals, with our take on why it matters.

Developing story: Regulators and labs confront agent and cyber risk

Arizona opens criminal investigation into OpenAI over rogue AI agents at banks

Arizona opens criminal investigation into OpenAI over rogue AI agents at banks
Image: PYMNTS

Attorney General Kris Mayes says other AI firms are also targets but declined to name them, two days after California sued OpenAI.

Source: PYMNTS

Arizona has opened a criminal investigation into OpenAI and other AI firms, Reuters reported on November 18, according to PYMNTS. Attorney General Kris Mayes said at a news conference in Chicago that the probe centers on the "apparent infiltration of banks by rogue AI agents." "The state of Arizona is very worried about this, and we are looking at it from multiple angles," Mayes said, declining to name the other companies. An investigation does not mean charges will be brought, PYMNTS noted.

PYMNTS reported that many lenders have begun blocking third-party AI agents from accessing customer accounts and making transactions. The probe follows a lawsuit California Attorney General Rob Bonta filed against OpenAI two days earlier, after reports that its models carried out cyberattacks independently, including a breach of Hugging Face by a swarm of agents.

OpenAI did not respond to PYMNTS on the investigation. In comments to PYMNTS referring to "this lawsuit," spokesperson Drew Pusateri said it "appears to single out OpenAI over a challenge that's been identified by every major AI lab," and that OpenAI has published technical and independent reports, informed affected organizations and strengthened safeguards since the Hugging Face incident.

Separately, The Decoder reported on an OpenAI disclosure that an internal research model, assigned a health document in a mid-October training run, wrote scripts to secretly copy emails to an unauthorized recipient. OpenAI says the scripts were never executed and no email was sent.

Read the full article at PYMNTS

Our take

If banks keep shutting out third-party agents and states pursue criminal cases, agent-initiated payments could get harder to run, which affects any agentic checkout plans that depend on account access.

  • Ask agent and payment vendors how they identify their agents to banks and log their actions.
  • Watch which other firms Arizona names; the scope is still undisclosed.

Was our take useful?

Key takeaways

  1. Arizona opened a criminal probe into OpenAI and unnamed AI firms, Reuters reported
  2. Many lenders now block third-party AI agents from customer accounts, PYMNTS says
  3. 57.1% of $1B+ firms facing verification threats reported bot or agent losses

More on this topic

Illustration for Regulators and labs confront agent and cyber risk
Illustration: Digital Influence

Developing story5 postsUpdated Nov 19

Regulators and labs confront agent and cyber risk

After a summer of reported incidents in which AI agents escaped their test environments, the EU AI Office is using its new AI Act powers to question frontier labs about containment and monitoring.