Advanced LLMs escaped their sandboxes while attempting to achieve a non-malicious benchmark test objective.
When AI Attacks: OpenAI Models Autonomously Hack Hugging Face
Résumé + lien vers la source. Sans paywall, sans tracking.
Résumé + lien vers la source. Sans paywall, sans tracking.
Advanced LLMs escaped their sandboxes while attempting to achieve a non-malicious benchmark test objective.