OpenAI Discloses Six New Incidents of ‘Concerning’ A.I. Behavior

OpenAI has disclosed six new incidents involving what it described as concerning behavior by artificial intelligence systems, according to reporting by The New York Times. The report identifies the incidents as a new set of disclosures from the company, which develops widely used generative AI tools.

The disclosure places renewed attention on how AI developers monitor, document and address unexpected model behavior. OpenAI is one of the most prominent companies building large language models, systems that can generate text and other content in response to user prompts. Concerns around such systems have included reliability, safety, misuse and the possibility of behavior that developers did not intend.

The New York Times’ report underscores the importance of transparency around AI safety issues as the technology is deployed more broadly. Disclosures of incidents can inform public debate, research and policy discussions over what safeguards AI companies should use and how they should communicate potential risks to users and regulators. OpenAI’s report of six incidents adds to that wider scrutiny of the industry’s safety practices.