Supporters Seek Coordinated Safeguards
Warnings about uncontrolled artificial intelligence have renewed proposals for a coordinated slowdown in development. Several executives support cooperation between companies and governments to prevent severe risks. However, industry leaders disagree about whether collective limits are necessary or whether individual companies can manage safety independently.
Anthropic CEO Dario Amodei offered the most detailed plan. He proposed giving independent evaluators continuous access to frontier laboratories, including offices, credentials, and company equipment. Anthropic and OpenAI have committed to this approach. Amodei also supports regulation and coordination among democratic governments, while acknowledging that cooperation with China would be difficult.
OpenAI Supports Slower but Continued Progress
OpenAI CEO Sam Altman favors pacing development and expanding cooperation across the industry. The company supports mandatory national safety requirements, stronger state laws, and common global standards. Its proposal includes shared testing, independent assessments, cybersecurity protections, incident reporting, and monitoring for systems that can improve themselves.
Altman emphasized that pacing does not mean stopping innovation. Development would remain rapid but proceed more slowly because safety evaluations and monitoring require time and resources. OpenAI argued that competitive pressure should not justify reckless decisions. Elon Musk also endorsed Amodei’s plan and proposed allowing rival companies to review new models before release.
Google DeepMind leader Demis Hassabis agreed with the direction of the proposal. He suggested an industry-funded standards body to develop testing protocols for national-security risks. The organization would need enough technical talent and computing resources to evaluate advanced systems effectively.
Meta and Nvidia Reject Collective Limits
Meta CEO Mark Zuckerberg opposed a coordinated slowdown and said each company should ensure its own models are safe. He argued that laboratories already face strong incentives and significant liability to prevent harm. Nvidia CEO Jensen Huang similarly said companies should adjust their own pace without new laws or regulations.
Microsoft leaders supported deliberate development but questioned control by a small group of frontier laboratories. CEO Satya Nadella called for broad representation from governments, academia, industries, and different countries. Microsoft AI chief Mustafa Suleyman shared the safety objective but warned that treating models like conscious beings could amplify risks. The debate therefore centers on who should set limits, supervise companies, and decide when development must slow.
References
Huamani, K. (2026, September 16). Divisions emerge in the tech industry over calls for a coordinated AI slowdown. AP News. https://apnews.com/article/ai-slowdown-anthropic-openai-meta-nvidia-1d9615931af28a83cb97489178e90f2d