Anthropic, the AI firm behind the Claude models, proposed a global pause on the development of the most advanced artificial intelligence systems on June 4, 2026. The company cited concerns that such AI could soon become uncontrollable and pose serious safety risks without coordinated oversight [1, 2, 3, 4, 5, 6, 7, 8, 9, 10].
The company said that an effective pause would require multiple leading AI companies, especially those in the US and China, to halt development simultaneously under verifiable international rules [1, 2, 4, 11, 6, 7, 8, 9, 10]. Anthropic warned that without such global coordination, competitive and geopolitical pressures will force companies and governments to make difficult safety decisions independently [1, 2, 6, 7, 8, 10].
Anthropic detailed the specific risk of AI systems achieving recursive self-improvement, enabling them to autonomously create increasingly capable successors. The company said this poses unprecedented challenges for governance and safety [4, 11, 5, 8, 9, 10]. Co-founders Jack Clark and Marina Favaro described it as "that collision, where recursive intelligence building itself ever faster meets the world of humans, relationships, and governance," which represents an unpredictable future [10].
The company compared the call for an AI development pause to nuclear arms control treaties but noted AI is harder to regulate due to the ease of hiding training runs and rapid innovation cycles [1, 4, 9, 10]. Jack Clark said, "We need to be able to choose to hit the gas pedal or the brakes. But right now, the AI industry is like a car with a gas pedal but no brake pedal" [11].
Anthropic plans to convene policymakers, researchers and AI companies to discuss global coordination mechanisms and will publish the outcomes of these discussions [4, 5, 9, 10]. Meanwhile, the company continues to develop and release advanced models like Claude and Mythos, the latter recognized by the White House for cybersecurity but not publicly available [1, 2, 4, 5, 6, 7, 9]. Anthropic is also preparing for an initial public offering [4, 9].
The company’s call echoes previous AI pause efforts, including a 2023 petition by the Future of Life Institute, signed by high-profile figures like Elon Musk, which requested a six-month pause on advanced AI to implement safeguards [4, 9, 10].
The response to Anthropic’s announcement has been mixed. White House and industry officials have pushed back, arguing the risks are overstated and that a pause could slow US innovation while giving China a strategic advantage in AI development [1, 2, 4, 11, 6, 7, 9]. U.S. President Donald Trump discussed potential US-China cooperation on AI safety during a recent visit and signed an executive order requiring a 30-day review of the most powerful US AI models before release [1, 2, 6, 7].
Anthropic had previously pledged to pause development if it lacked a significant lead but loosened that stance earlier in 2026 [10]. The company emphasized, "We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology" [1].
