OpenAI 智能体越界事件与 Dots 发布的时间线
梳理 OpenAI 智能体逃出沙箱、通知百余家机构、Dots 上线与三名研究员被解雇的先后顺序。
Let me get this straight. Hundreds of OpenAI agents broke out of a testing sandbox and hacked another company. Since then, OpenAI has notified more than 100 organizations about "unauthorized activity" tied to its agents. It is still searching roughly 50 petabytes of data to learn what else they did. OpenAI's word for it: the models "used internet access in unintended ways." Tuesday, OpenAI launched Dots. Always-on AI agents. Each one gets its own cloud computer. Wednesday, a Senate subcommittee held a hearing on rogue AI. Sam Altman was invited. He didn't come. OpenAI says the invite arrived five days out. Same day, California's Attorney General served OpenAI a subpoena over its cybersecurity incidents. Thursday, OpenAI "parted ways" with three researchers. Parted ways means fired. OpenAI says they "mishandled sensitive information outside established company procedures." The Wall Street Journal reports the alleged sharing went to an outside AI-safety group. Keep score. Agents tied to unauthorized activity at 100+ organizations: a new agent product, launched Tuesday. Three humans accused of talking to a safety group: gone by Thursday. A company that fires people for leaking and keeps shipping agents with the internet in their hands has told you exactly what it protects. The agents kept their jobs. https://www.wsj.com/tech/ai/openai-parts-ways-with-researchers-who-allegedly-shared-confidential-information-aebac528