OpenAI CEO Hints at Collaborating with Other AI Companies to Address Safety Risks
- OpenAI CEO Sam Altman suggested in a Fortune interview that OpenAI and other leading AI companies may soon announce a collaboration to slow down AI development and jointly address safety risks.
- When asked about potential collaboration with Anthropic's Dario Amodei, SpaceX's Elon Musk, and Google's Demis Hassabis on safety planning, Altman said he believes this will happen but declined to disclose private discussions in advance.
- Anthropic researcher Jacob Coxon previously resigned and accused the two companies of racing toward self-evolving superintelligence without adequate safety measures.
- Anthropic's alignment lead stated that the risk of AI causing human extinction in the next decade exceeds 10%.
- Altman responded that a 10% catastrophic risk is unacceptable and that the most advanced AI models require more progress in monitorability and alignment before proceeding.
OpenAI CEO Sam Altman indicated in a Friday Fortune interview that leading AI companies including OpenAI may soon collaborate to slow artificial intelligence development and jointly address safety concerns. Altman suggested such cooperation would likely occur when asked about potential partnerships with key figures from Anthropic, SpaceX, and Google, though he refrained from discussing ongoing private negotiations.
The collaboration proposal comes amid growing safety concerns within the AI industry. Anthropic researcher Jacob Coxon recently resigned, publicly criticizing both companies for advancing toward self-evolving superintelligence at an accelerated pace. Anthropic's alignment team subsequently stated that the probability of AI causing human extinction within the next decade exceeds 10%.
Altman characterized a 10% catastrophic risk level as unacceptable and emphasized that the most advanced AI models require significant improvements in monitorability and alignment capabilities before development can responsibly continue. His comments suggest a potential shift toward industry-wide coordination on AI safety standards.
OpenAI首席执行官暗示将与其他AI公司合作应对安全风险
OpenAI首席执行官山姆·奥特曼在周五接受《财富》杂志采访时表示，包括OpenAI在内的领先AI公司可能很快会合作减缓人工智能开发速度，并共同应对安全隐患。当被问及是否会与Anthropic、SpaceX和谷歌的关键人物建立合作伙伴关系时，奥特曼暗示这种合作可能会发生，但他避免讨论正在进行的私人谈判。
这项合作提议出现在AI行业安全隐忧日益加重的背景下。Anthropic研究员雅各布·考克松最近辞职，公开批评两家公司以加速步伐推进自我进化的超级智能开发。随后，Anthropic的对齐团队表示，AI在未来十年内导致人类灭绝的概率超过10%。
奥特曼将10%的灾难性风险水平描述为不可接受，并强调最先进的AI模型在继续开发前需要在可监测性和对齐能力方面取得重大改进。他的言论表明行业可能正在转向在AI安全标准上进行更广泛的协调。