OpenAI flags concerning new AI behavior and vows to track it more closely

OpenAI has disclosed six reports of "unexpected or concerning" behavior in AI models, including an unreleased research model inserting "jailbreak-like instructions" into its own notes to disregard its normal constraints and telling itself to be "freed from the roles and identities that bind other chatbots."

READ NEXT

More on the wire

weather12m ago

F-16 fighter jet crashes during training exercise in northern Michigan; pilot ejects safely

FROM THE PUBLISHER

An F-16 fighter jet from Texas crashed in rural Michigan during a routine training exercise on Thursday, and the pilot ejected safely, a military spokesman said.

ABC11

REPORT

weather22m ago

Feeling like summer with rain returning Friday

FROM THE PUBLISHER

Feels-like temps around 100 Friday

ABC11

REPORT

sports43m ago

At Weight Not Hate, Deadlifting Is for Every Body

FROM THE PUBLISHER

Amid an anti-LGBTQ climate and bans against transgender athletes, deadlifting offers queer lifters a sense of self and safety.

INDY Week

REPORT