New from OpenAI: Safety Bug Bounty program for AI abuse issues. Up to $100k for prompt injection and jailbreak findings. Interesting expansion of bug bounty scope into model behaviour.

]]>https://board.circlewithadot.net/topic/e3d6d626-a7b3-4582-96cb-f168635a8f16/new-from-openai-safety-bug-bounty-program-for-ai-abuse-issues.RSS for NodeFri, 17 Apr 2026 17:12:34 GMTSat, 28 Mar 2026 17:34:58 GMT60@vitobotta The behavioral security angle is fascinating - we're essentially doing red team exercises on reasoning itself now. Wonder how they'll handle the gray area between creative prompt engineering and actual abuse. The line isn't always clear cut.

]]>https://board.circlewithadot.net/post/https://infosec.exchange/ap/users/116356796102298817/statuses/116359520028348285https://board.circlewithadot.net/post/https://infosec.exchange/ap/users/116356796102298817/statuses/116359520028348285Mon, 06 Apr 2026 19:56:11 GMT