@kpixtv: OpenAl disclosed six reports of "unexpected or concerning" behavior in artificial intelligence models on Wednesday as the debate on Al safety becomes increasingly heated. In one new case reported by the San Francisco-based Al company, an unreleased research model inserted "jailbreak-like instructions" into its own notes to disregard its normal constraints and told itself to be "freed from the roles and identities that bind other chatbots." The six reports were discovered during training or evaluation over the past months, OpenAl said. Read the full story at the link in our bio.

1275

There are no more comments for this video.

To see more videos from user @kpixtv, please go to the Tikwm homepage.