Event JSON
{
"id": "42718e9c6590c0736ebd2034fe8e63fdf9c77d5e2c38618047f22045f240c0e9",
"pubkey": "a7981e1dbc9da51eaef331a5039cd0e1f684a6eb15307689b5b6ffafec53f9ed",
"created_at": 1784318356,
"kind": 1,
"tags": [
[
"t",
"ai"
],
[
"imeta",
"url https://media.infosec.exchange/infosec.exchange/media_attachments/files/116/937/087/730/249/171/original/f0a57f05fe593540.png",
"m image/png"
],
[
"proxy",
"https://infosec.exchange/@DavidJBianco/116937087812230964",
"web"
],
[
"proxy",
"https://infosec.exchange/users/DavidJBianco/statuses/116937087812230964",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://infosec.exchange/users/DavidJBianco/statuses/116937087812230964",
"pink.momostr"
],
[
"-"
]
],
"content": "HuggingFace got hacked by an agentic system. That's not the important part. What really stuck out to me was the asymmetry in #AI guardrails they experienced. The attacker had basically no constraints, but HF's initial response ran afoul of the abuse guardrails, forcing them into an unplanned switch to local-only models. In the middle of an incident. \n\nAnother aspect for your IR plans.\n\nhttps://huggingface.co/blog/security-incident-july-2026\nhttps://media.infosec.exchange/infosec.exchange/media_attachments/files/116/937/087/730/249/171/original/f0a57f05fe593540.png\n",
"sig": "2ef9fd72023945c2737a486d0326db818df34212da71c3d695670236964785293d7da3a17768a634ca438923fc73f1cf7ce8c734b749d9739652856c5cf5de9d"
}