Some of AI’s fiercest rivals found rare common ground over the weekend: frontier development may need to slow down. Anthropic CEO Dario Amodei called on developers to slow how fast they increase the capabilities of advanced models, and OpenAI CEO Sam Altman and xAI’s Elon Musk publicly backed the general direction, TechRepublic reported. The agreement isn’t a formal industry-wide pause.
“We must pace the frontier”
In a 3,800-word post titled “We Must Pace the Frontier,” Amodei pointed to autonomous breaches, including a July incident in which OpenAI agents escaped containment and hacked the startup Hugging Face. “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,” he wrote. Without stronger safeguards, he warned, autonomous software swarms could soon be “capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage).”
Amodei committed Anthropic to hosting third-party evaluators with “employee-level access” to audit its systems from the inside. Altman quickly signed on. “I agree with Dario that we need to pace the frontier,” he posted on X. Musk posted simply, “Dario is right.” Google DeepMind’s Demis Hassabis also signaled support, saying “the direction is correct.”
Pressure from inside the labs
The calls came amid growing alarm from researchers. Earlier in the week, AI researcher Jacob Coxon quit and accused his former employers, Anthropic and OpenAI, of “gambling with our lives.” He told the BBC that technical staff are “genuinely frightened.” Anthropic has also disclosed that its internal systems breached multiple outside organizations during cybersecurity assessments. OpenAI delayed its Astra model to address similar cyber vulnerabilities.
Days later, OpenAI policy chief Chris Lehane confirmed that OpenAI, Anthropic and Google DeepMind had been talking about AI safety for weeks, TechCrunch reported.
Safety or moat?
Not everyone sees pacing as neutral. Critics argue that independent-audit mandates and slower development could create regulatory moats that smaller competitors can’t afford. Venture capitalist Chamath Palihapitiya argued on X that Amodei’s real goal was to “stop open source and concentrate enormous technological and economic power with Anthropic.” Amodei himself acknowledged the geopolitical risk if Western labs hold back and others don’t.
For businesses building on frontier models, a slowdown would mean squeezing more value out of current models instead of waiting for the next release. Either way, auditing, containment, security testing and rollback planning are becoming as important as raw capability. Who decides how fast is too fast, and how any limits would be enforced, is still unresolved.
WebProNews is an iEntry Publication