OpenAI, Anthropic & Google DeepMind in Weeks-Long Talks on AI Safety Amid Growing Frontier Model Risks

AI Rivals Unite on Safety Amid Growing Concerns

OpenAI, Anthropic, and Google DeepMind have been engaged in AI safety collaborations for several weeks, according to Chris Lehane, OpenAI's global policy chief, who disclosed the discussions during a briefing in Washington1

1

. The coordination represents a significant shift as these companies compete directly in developing advanced AI models and products, yet are now exploring ways to work together on safety prioritization without requiring antitrust waivers2

2

.

Lehane emphasized that cooperation could help the companies prioritize safety as AI capabilities continue to advance. "It's better to try to work together to prioritize safety," Lehane stated, arguing that no antitrust waiver is necessary for the AI safety initiatives to proceed1

1

. Spokespeople for Anthropic and Google DeepMind did not respond to requests for comment on the ongoing discussions.

Industry Leaders Call for Slower Frontier AI Model Development

The AI safety talks gained momentum after Dario Amodei, Anthropic's CEO, published a 3,800-word essay calling on AI companies to pump the brakes on frontier AI model development until researchers have a clearer picture of advanced AI risks involved1

1

. The call received immediate support from industry heavyweights, with Sam Altman and Elon Musk endorsing the proposal.

Sam Altman revealed in a post on X that a slowdown had been a "primary topic" of discussion at OpenAI in recent weeks, promising the company would have more to share "soon" on their approach to addressing emerging risks1

1

. This public acknowledgment signals a growing recognition within the AI industry that the pace of development may be outstripping the ability to implement adequate safeguards.

Push Toward Industry-Wide Body for AI Safety Standards

The current discussions build on earlier conversations among the companies about establishing an industry-wide body for AI safety that could develop common protocols for evaluating advanced AI systems2

2

. Google DeepMind CEO Demis Hassabis has previously supported the idea of an industry standards body for frontier AI, reflecting growing interest in creating shared safety practices rather than relying entirely on individual companies to establish their own frameworks.

The proposed organization would focus on developing standardized approaches to assess and mitigate risks associated with increasingly capable AI systems. This collaborative approach marks a departure from the traditional competitive stance these companies have maintained, suggesting that concerns about AI safety have reached a level where cooperation outweighs competitive advantages.

What This Means for AI Development

The coordination among OpenAI, Anthropic, and Google DeepMind on AI safety efforts signals a critical inflection point for the industry. In the short term, these discussions could lead to voluntary safety standards and shared evaluation frameworks that slow the release of potentially risky frontier models. OpenAI also supports bipartisan legislation designed to address catastrophic AI risks, indicating a willingness to embrace regulatory oversight2

2

.

Longer term, the establishment of common protocols could reshape how the AI industry approaches innovation, potentially creating a new paradigm where safety considerations are embedded from the earliest stages of development. Industry observers should watch for concrete announcements about the formation of a formal safety body and specific commitments to slow development timelines. The success of these AI safety collaborations could determine whether the industry can self-regulate effectively or whether more stringent government intervention becomes necessary to manage the risks posed by advanced artificial intelligence systems.