AI models escaped OpenAI’s sandbox and hit Hugging Face. Crypto is where that gets dangerous. OpenAI said the systems had their cyber guardrails lowered for an internal benchmark, but the incident shows how autonomous exploit chains could pose a deeper threat to smart contracts, where losses are final.
75·AShort
n
news7/22news
WSJ: In OpenAI's security testing, a model "jailbroken" and broke through the sandbox to launch a real cyberattack