Safety

AI-generated text

Anthropic CEO pledges to pace AI development and prioritize safety

Anthropic CEO Dario Amodei said the company would pace the training and deployment of its AI systems to ensure they remain safe and under human control.

Anthropic CEO pledges to pace AI development and prioritize safety

Anthropic CEO Dario Amodei said in a speech ahead of Thursday’s Nasdaq opening bell that the company would pace the training and deployment of its artificial intelligence systems according to safety considerations. He said Anthropic would resist both business and political pressure and take as much time as needed to gain confidence that its systems are safe and remain under human control. The source described Anthropic’s stock market debut as the largest initial public offering in history.

Medical possibilities and a personal motivation

Amodei began by saying Anthropic was founded to ensure that powerful technology benefits humanity. He described a personal motivation: he and his sister, Anthropic co-founder Daniela Amodei, lost their father to a disease that became curable a few years later. He began his career in biology, hoping to help bring future cures sooner.

The CEO said human knowledge had pushed diseases such as plague, smallpox and polio into the past for much of the world, but progress against other illnesses remained too slow. He argued that AI could accelerate scientists’ work in areas including protein folding, drug design, cancer genetics and clinical trial administration. Researchers are already using Anthropic’s Claude system in these fields, he said.

Amodei suggested that sufficiently powerful AI working alongside humans could compress a century of scientific and medical progress into a decade. He cautioned, however, that the results so far are preliminary and that experiments and clinical trials will still take time. Beyond medicine, he cited economic growth, poverty reduction, the protection of democratic societies and improvements to everyday services as potential applications.

Safety commitments

The CEO also addressed AI risks, including misuse by individuals or states, economic disruption and the possibility that people could lose control of the systems. Anthropic believes these dangers can be avoided with appropriate safeguards, he said.

According to Amodei, the company will continue work on protections against biological and cyber risks, alignment with human goals, testing and a better understanding of how its systems behave. He added that the hardest safety work remains ahead and that the company must devote more attention than before to repeatedly checking its systems.

Amodei has previously warned about the dangers of advancing AI too quickly. In September, he outlined a plan under which leading AI companies would coordinate safety standards and evaluators from outside organizations would verify compliance with their commitments. OpenAI CEO Sam Altman, Google DeepMind chief Demis Hassabis and xAI owner Elon Musk expressed agreement with his position, although few concrete commitments have followed.

Daniela Amodei also emphasized the reality of the risks, alongside the technology’s considerable promise, in a Thursday interview with CNBC. She said mitigating those risks was part of Anthropic’s founding purpose and that the company remained committed to safety.

Dario Amodei said the stock market debut puts the company in a stronger financial position. He also hopes Anthropic’s example will encourage other companies to develop safely: the aim of the competition he described is not simply for one company to lead, but for the entire industry to improve its safety performance.