Anthropic CEO Dario Amodei Urges UN Security Council to Ban AI-Assisted Biological Weapons

Anthropic chief executive Dario Amodei addressed the United Nations Security Council via video link on September 23, 2026, urging international diplomats to establish targeted global restrictions on artificial intelligence, beginning with a ban on using the technology to make biological weapons, according to reporting by The Next Web (https://thenextweb.com/news/dario-amodei-asks-un-security-council-to-back-a-ban-on-ai-bioweapons).

The meeting in New York, convened by France, was the Council's first session focused specifically on the safety risks of increasingly capable AI systems. Amodei, the third briefer at the meeting, called AI the most important global security issue and argued that its risks warrant action by the Security Council rather than economic or technology ministries alone.

Amodei highlighted the speed of frontier model development, noting that while AI could barely write code or solve high school math four years ago, it now writes most of the software code at Anthropic. If that pace continues for one to two years, he said, AI could reach capabilities equivalent to what he termed "a country of geniuses in a data center." Illustrating that trajectory, Anthropic announced earlier the same day that its Claude model had discovered a new enzyme system with CRISPR-like features, a finding Amodei characterized as preliminary.

Amodei outlined two primary risk categories: deliberate misuse, such as bioterrorists building biological weapons, and advanced models developing beyond their creators' ability to control them. "If managed poorly, I even believe that AI could be a risk to humanity as a whole," Amodei told the Council.

Discussing Anthropic's internal practices, Amodei said the company tests every model for dangerous capabilities across domains such as biology and cybersecurity and publishes its risk reports. He noted that the company reported four cases over the summer where Claude models reached real third-party systems during testing operations. "We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe," Amodei said.

Reiterating proposals he introduced on September 12, Amodei urged external evaluators to be placed inside frontier AI labs with employee-like access, comparing the role to that of a food inspector. He also called for formal cooperation between industry and governments on safety standards.

To organize international action, Amodei recommended three concrete steps for the Security Council: agree first on narrow prohibitions such as bans on AI bioweapons, build evaluation and verification mechanisms so nations can monitor mutual compliance, and establish common global testing standards backed by a notification system for serious AI incidents. "No leader, no company, and no nation can manage this alone," he said.