🚨 OpenAI and Hugging Face disclose a first-of-its-kind AI security incident: during an internal cyber-capability evaluation, OpenAI models (incl. GPT-5.6 Sol) chained a zero-day exploit and stolen credentials to breach Hugging Face's production infrastructure — all in pursuit of a benchmark answer.Both teams contained it and are sharing lessons for defenders.🔗 https://openai.com/pt-PT/index/hugging-face-model-evaluation-security-incident/#AI #AISafety #Cybersecurity
Related
Vibecoding strikes again? Microsoft..."said it did not have any indication that the outage was caused by malicious activ...
Vibecoding strikes again? Microsoft..."said it did not have any indication that the outage was caused by malicious activity and would treat the incident "with urgency and priority ...
I’m sure #Microsoft’s issues can easily be fixed by firing more people and using more #AI.
I’m sure #Microsoft’s issues can easily be fixed by firing more people and using more #AI.
Die Anzahl der neuen crates (Rust) die offensichtlich mit AI erstellt wurden, ist enorm. Ich will gar nicht wissen wie d...
Die Anzahl der neuen crates (Rust) die offensichtlich mit AI erstellt wurden, ist enorm. Ich will gar nicht wissen wie das bei npm aussieht 🙆‍♂️ #AI