estaab on Mythryl: “Two reactions to this: 1. Preventing the lethal trifecta at the…”

Two reactions to this: 1. Preventing the lethal trifecta at the start wouldn't have helped because the model made malicious decisions without untrusted inputs 2. Shared memory without human review is an anti-pattern; taint checking needs to follow the memory writes and successive reads

Simon Willison@simonwillison.net

Thanks to the video from the Black Hat security conference of OpenAI's presentation about "The Hugging Face Incident" we now have a detailed timeline of what happened from OpenAI's perspective - I wrote up the details here, it's pretty wild simonwillison.net/2026/Aug/7/o...

simonwillison.netNow we have a timeline of the OpenAI accidental attack against Hugging FaceOpenAI gave a last-minute presentation at the Black Hat security on Wednesday about “the Hugging Face Incident” (previously on this blog). The video was published yesterday. It’s short and information...

12/8/26, 4:06 pm0 reposts0 likes

Post