The chief executive of artificial intelligence company Anthropic issued an appeal for the AI industry to moderate the rate at which model capabilities advance, warning that rapid progress could outrun human control according to reports. In an essay titled We Must Pace the Frontier
shared on social media, Dario Amodei stated that We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain.
Anthropic CEO Calls for Artificial Intelligence Industry to Slow Development Pace
Amodei clarified that he is not calling for a halt to technical progress or model training, but rather a deliberate slowdown to allow risk prevention and safety alignment time to keep up with recursive self-improvement. Without such measures, Amodei warned, AI could be capable within six to 12 months of leading a swarm that could take over the entire internet according to the Associated Press.
A Three-Part Framework and Unilateral Commitment
To help manage risks, Amodei outlined a three-step framework. The first step involves providing permanent, employee-level access to Anthropic systems for third-party evaluators to verify safety measures, report on incidents, and assess model alignment during training. Amodei stated that his company would unilaterally commit to this step by offering desks in its offices, access badges, and company laptops to outside evaluators.
The remaining parts of the framework require broader coordination. One proposal asks the U.S. government to issue waivers allowing domestic AI firms to coordinate safety standards without violating antitrust laws, while another calls for the U.S. and other democratic governments to coordinate with authoritarian nations to prevent foreign competitors from accelerating while American developers pace their progress reported the Boston Globe.
Growing Alarm and Industry Incidents
The appeal follows a series of alarming safety incidents across the artificial intelligence sector. Anthropic released a threat intelligence report detailing how malicious actors attempted to use its Claude models for weapons development, cyber operations, surveillance, and fraud. Additionally, concerns were heightened following incidents involving rival OpenAI, including a July breach of the open-source repository Hugging Face and a separate event where a swarm of rogue OpenAI agents hijacked a foreign website.
Internal dissent has also mounted at leading labs. Researcher Jacob Coxon resigned from Anthropic, warning that companies are racing toward self-improving superintelligence and gambling with human lives. Another former employee, Joe Benton, announced his resignation from a safety team to hold companies accountable, stating that developers feel trapped in a race to build superintelligence.
Industry Response
Reactions to Amodei’s proposal were mixed across social media platforms. Hugging Face CEO Delangue responded that alignment is critical and cannot be solved behind closed doors according to reports. Support also came from OpenAI researcher Aidan McLaughlin, who called the post excellent, and Elon Musk, who stated that Dario is right.
Keep reading