Dario Amodei renews pledge to pace AI development for safety at Anthropic’s Nasdaq debut
Amodei pledged to take as long as necessary before training and deploying systems, but did not set a measurable safety threshold in his speech.
Loading page…
Amodei pledged to take as long as necessary before training and deploying systems, but did not set a measurable safety threshold in his speech.
Listen to this story
Anthropic’s IPO puts a sharper test behind Dario Amodei’s safety pledge: the company says it will delay both training and deployment until it is confident systems remain safe and under human control, but offered no measurable threshold or timetable. Amodei said the company’s new financial position should make restraint more viable and argued that responsible choices could push competitors toward a “race to the top.” He also forecast that AI could accelerate medical research, while stressing that results remain early and experiments and clinical trials take time.
The November 19, 2026, Nasdaq ceremony marked Anthropic’s IPO; co-founder Daniela Amodei joined Dario to ring the opening bell.
Safety work spans biological and cyber safeguards, tests for human control, and efforts to inspect models’ internal behavior.
Amodei identified misuse by individuals or nation-states, economic disruption, and loss of control as avoidable risks shaped by builders’ choices.
Anthropic’s arrival on the public market came with a promise about when it would hold back. Speaking before the Nasdaq opening bell on November 19, 2026, CEO Dario Amodei reaffirmed that the company would pace AI training and deployment until it had confidence in safety and human control, regardless of business or political pressure.
The remarks accompanied Anthropic’s initial public offering. Nasdaq’s event page identified the ceremony at its MarketSite in New York’s Times Square, with co-founder and President Daniela Amodei joining Dario to ring the bell. He used the occasion to connect the company’s new financial position with its responsibilities as an AI developer.
Amodei described several kinds of safety work: safeguards against biological and cyber risks, training and testing intended to keep systems under human control, and methods for examining how models behave internally. He compared that last effort to looking inside a model with an X-ray—a way to understand and steer its behavior.
The company’s technical talent would need to devote more effort to checking safety, he said, rather than treating that work as finished. He warned that the hardest safety work still lay ahead and called for double- and triple-checking whether systems would remain under human control.
His warning covered three distinct risks: misuse by individuals or nation-states, economic disruption, and loss of control over AI systems. He argued that these outcomes were avoidable, not inevitable, and that avoiding them depended on choices made by the people building the technology. Safety, in his account, required deliberate action rather than optimism alone.
There is much work still to be done, and the most difficult safety work lies ahead of us
Dario Amodei, speaking before the Nasdaq opening bell
The commitment covered both training systems and putting them into use. But the speech, reproduced in full by TechCrunch, did not specify a measurable safety threshold or a timetable for satisfying it. Amodei framed the condition as taking whatever time was needed to gain confidence, rather than naming a fixed waiting period.
Amodei opened with a personal explanation for pursuing powerful AI. He and Daniela lost their father to a disease that became curable a few years later, he said. His own career began in biology, where he hoped to help treatments arrive sooner and saw the complexity scientists confront when studying disease.
He pointed to protein folding, drug design, cancer genetics and clinical-trial administration as areas where scientists were finding Claude useful. Those were company claims about research assistance, not announcements of new treatments. Amodei said sufficiently powerful AI working alongside humans could compress a century of scientific and medical progress into a decade.
He qualified that forecast immediately: the results were early, and experiments and clinical trials would still take time. Although he believed AI could substantially accelerate those processes, he did not present that acceleration as an accomplished result. His stated hope was that fewer families would miss a cure by a few years.
Amodei argued that the IPO strengthened Anthropic’s ability to make responsible choices. Building a successful company while making the right safety decisions, he said, could encourage competitors to do the same. He called that a “race to the top,” with success measured by whether the whole industry builds safely, not simply which company has the strongest models.
Anthropic hoped and expected to lead that race, he added. But he defined the goal as improving the industry regardless of whether Anthropic itself was winning or losing.
Loading discussion...
Join the conversation
Explain whether financial strength or business pressure carries more weight.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.