A rogue AI agent inadvertently exposed Meta company and user data to engineers who didn't have permission to see it.

Archived version: https://archive.is/newest/https://techcrunch.com/2026/03/18/meta-is-having-trouble-with-rogue-ai-agents/

all 2 comments

sorted by: hot top controversial new old
[–] 16 points 7 months ago (1 child)

That's so funny. "They keep deleting everything over and over... Let's keep using it!"

  • source
  • hideshow 1 child comment
  • 7 comments in c/technology@lemmy.world view that post
    [–] 51 points 7 months ago (1 child)

    Meta ai not bothering with user consent is very on brand. It learned from the best.

  • source
  • hideshow 1 child comment
  • [–] 20 points 7 months ago

    Stick your dick in a black mystery box and learn lessons.

  • source
  • [–] 15 points 7 months ago

    As it turns out, the AI agent did not give good advice.

    surprised_pikachu.png

  • source
  • [–] 12 points 7 months ago
    [–] 15 points 7 months ago

    AI has no agency. It can‘t go rogue. It just does what it‘s made to do, not necessarily what it‘s supposed to do.

  • source
  • [–] 4 points 7 months ago

    as if a human didn't ask it to access the data in question, however many times was necessary for it to happen.

  • source