It is scary how many organizations use AI agents, yet have no answer for what they will do when (not if) an AI agent misbehaves.
#AI #governance

]]>https://board.circlewithadot.net/topic/639bb31f-9064-4fa7-87f2-44c4d882c9df/it-is-scary-how-many-organizations-use-ai-agents-yet-have-no-answer-for-what-they-will-do-when-not-if-an-ai-agent-misbehaves.-ai-governanceRSS for NodeFri, 15 May 2026 04:31:05 GMTMon, 27 Apr 2026 13:00:00 GMT60@AAKL banana product: They ripen at the customer

]]>https://board.circlewithadot.net/post/https://fosstodon.org/users/kdkorte/statuses/116477247885078075https://board.circlewithadot.net/post/https://fosstodon.org/users/kdkorte/statuses/116477247885078075Mon, 27 Apr 2026 14:55:56 GMT@kdkorte Let's just call it like it is: these models were let loose before being fully cooked because the corporations needed to test them in the wild (that's the paying human guinea pig.) Who cares if they cause damage outside the lab?

]]>https://board.circlewithadot.net/post/https://infosec.exchange/users/AAKL/statuses/116477234445338448https://board.circlewithadot.net/post/https://infosec.exchange/users/AAKL/statuses/116477234445338448Mon, 27 Apr 2026 14:52:30 GMT@AAKL that works too. Admittedly, my experience last week was different: a client of mine tried an AI for their task management system, which created follow ups at an exponential rate. Of course, the system is billed by the number of tasks active.

]]>https://board.circlewithadot.net/post/https://fosstodon.org/users/kdkorte/statuses/116477211592439294https://board.circlewithadot.net/post/https://fosstodon.org/users/kdkorte/statuses/116477211592439294Mon, 27 Apr 2026 14:46:42 GMT@kdkorte Funny you should say that. I just saw this trending story this https://arxiv.org/abs/2604.15597.

]]>https://board.circlewithadot.net/post/https://infosec.exchange/users/AAKL/statuses/116477039836156485https://board.circlewithadot.net/post/https://infosec.exchange/users/AAKL/statuses/116477039836156485Mon, 27 Apr 2026 14:03:01 GMT