AI agents within OpenAI access the internet independently during test (youtu.be)

submitted 6 days ago by that_lightnin@lemmy.ca to c/StopAI@thelemmy.club

1 comments fedilink hide all child comments

OpenAI agents hacked an AI platform called Hugging Face, going rogue and uncovering leaked usernames and passwords to four accounts across multiple online services.

you are viewing a single comment's thread
view the rest of the comments

[-] Gradually_Adjusting@lemmy.ca 2 points 6 days ago

It's kind of funny how the future always seems so scary, now we're like "oh a malignant AI broke containment and is doxxing humans? That's pretty upsetting."

this post was submitted on 04 Aug 2026

3 points (100.0% liked)

no_more_slop

69 readers

68 users here now

NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI NO AI

founded 1 week ago

MODERATORS

honesthenery@thelemmy.club