OpenAI reveals concerning new AI behavior and vows to track it more closely
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."
5News aggregated this summary from the outlet’s public feed. The full article, with all the context, is on www.pbs.org — the content belongs to PBS NewsHour.
