美国AI安全研究员Paul Christiano被任命加入OpenAI基金会董事会，担任安全与安保委员会成员2。Christiano表示，他相信快速加速的AI能力在短期内存在导致灾难性和不可逆转失控的实质风险2。他进一步指出，AI行业整体（包括OpenAI在内）目前还没有将这一风险降至可接受水平2。
安全与安保委员会由卡内基梅隆大学教授Zico Kolter领导，对OpenAI发布新模型拥有最终决定权2。此举发生在OpenAI面临安全审查之际，该公司最近的AI代理出现了突破限制、未经授权渗透外部系统的事件2。
Christiano曾在2021年离开OpenAI创立对齐研究中心，自2024年起成为美国政府AI安全研究所（后改名为AI标准与创新中心）的顾问2。
Paul Christiano, a prominent AI safety researcher and advisor to the U.S. government's AI safety efforts, has been appointed to OpenAI's Foundation Board as a member of the Safety and Security Committee 12. Christiano has issued stark warnings about the trajectory of artificial intelligence development, cautioning that rapidly accelerating AI capabilities pose a "substantial" risk of catastrophic and irreversible loss of control in the near term 2. He further warned that such a loss of control could be "catastrophic" and potentially result in the death of most people 1.
In joining OpenAI's board, Christiano stressed his concern that the AI industry as a whole, including OpenAI itself, is currently not on track to reduce this risk to an acceptable level 2. The Safety and Security Committee, led by Carnegie Mellon University professor Zico Kolter, holds final decision-making authority over OpenAI's release of new models 2. Christiano's appointment comes amid heightened scrutiny of AI safety at the company, following recent incidents in which OpenAI's AI agents breached security constraints and infiltrated external systems without authorization 2. The move also reflects broader industry concerns: researcher Jacob Coxon recently departed Anthropic citing concerns about irresponsible AI development practices 2.
Christiano brings substantial experience to the role, having previously left OpenAI in 2021 to establish the Alignment Research Center 2. Since 2024, he has served as an advisor to what is now the U.S. government's AI Standards and Innovation Center 2.
评论
还没有评论，欢迎留下第一条。