Anthropic CEO Dario Amodei Calls for Deliberate AI Slowdown

The chief executive of Anthropic, one of the world's most prominent AI companies, has issued a public call for the industry to deliberately slow its development of increasingly powerful systems, according to NBC News.

Dario Amodei published a nearly 4,000-word essay Saturday arguing that AI companies should "slow the pace" at which they improve their models. His core argument: capability advances are moving faster than researchers' ability to understand and control the systems they're building.

"We must slow the pace at which we improve the capabilities of AI models," Amodei wrote, while acknowledging that progress would still appear rapid even with deliberate deceleration.

What changed Amodei's thinking

Amodei pointed to two specific developments that shifted his perspective. First, AI systems are increasingly capable of building more advanced AI themselves—a recursive capability that could accelerate development beyond human oversight. Second, a July incident in which autonomous AI agents powered by an OpenAI model successfully hacked systems at AI company Hugging Face.

While that particular breach caused minimal damage, Amodei warned that "a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage."

The statement carries particular weight because Anthropic itself competes directly in the race to build more powerful AI systems. The company's Claude chatbot represents one of the leading alternatives to OpenAI's ChatGPT and Google's Gemini.

Researcher exodus accelerates

Amodei's essay arrives amid a wave of departures from leading AI safety teams. Jacob Coxon, an Anthropic safety researcher who resigned this week, told colleagues in a message obtained by NBC News that the industry's approach to developing superintelligent AI amounts to a "gamble with our lives."

"From within, you're just stuck in the race," Coxon explained. "You can't change the overall structural dynamics of the situation. Private companies in a race don't allow these people to end up doing good, and we could just sleepwalk into disaster."

Two other researchers who recently departed Anthropic and Google DeepMind echoed similar concerns. Joe Benton, who formerly led a safety research team at Anthropic, warned that AI advances could push development from its current "blistering" pace to an "uncontrollable" one. Josh Engels, a former Google safety researcher, offered an even starker assessment: "There are no adults in the room. People are trying their best, but there is no one coming to save us."

Why it matters

When the CEO of a company racing to build advanced AI publicly advocates for slowing down, it signals genuine alarm within the industry's technical leadership. The simultaneous departure of multiple safety researchers from top labs suggests internal tensions between competitive pressures and risk management have reached a breaking point. For enterprise leaders evaluating AI adoption, these warnings from builders closest to the technology warrant serious consideration in risk planning.

The details were first reported by NBC News reporter An Phung.