Archive · 24 Sep 2026
Anthropic CEO Warns UN of AI Risks to Humanity
Dario Amodei leads a company called Anthropic that builds artificial intelligence.
He told the United Nations Security Council that AI is improving very quickly.
He said this could help people with medicine, education, science, and work.
But he warned that people might misuse AI to make biological weapons.
He also said very advanced AI might become difficult for humans to control.
Anthropic tests its systems and adds extra safety protections.
Amodei said one company cannot solve these problems alone.
He wants countries and technology companies to create shared safety rules and cooperate worldwide.
Anthropic CEO Dario Amodei told the United Nations Security Council that rapidly advancing AI could threaten humanity if poorly managed.
He identified misuse for biological weapons and loss of control over increasingly capable AI systems as two major risks.
Amodei said AI could soon perform the work of many highly skilled professionals, while also producing benefits such as scientific discoveries.
Anthropic tests its models for dangerous capabilities, applies stronger safeguards, publishes safety research, and may slow development if needed.
Amodei called for cooperation among AI companies and governments, including global safety standards and a ban on AI-created biological weapons.
- Who
- Dario Amodei of Anthropic and Sam Altman of OpenAI addressed AI risks.
- What
- They warned that rapid AI development could enable dangerous misuse or cause humans to lose control of advanced systems.
- When
- Amodei and Altman delivered their warnings on the same Wednesday.
- Where
- The warnings were made before the United Nations Security Council.
- Why
- They called for international cooperation because AI risks could affect people worldwide and cannot be managed by individual companies alone.
AI Benefits and Progress
AI Risks and Controls
Rapid development
AI Benefits and Progress
Faster AI progress could produce major advances in medicine, education, productivity, and scientific discovery.
AI Risks and Controls
Uncontrolled progress could enable biological weapons or move beyond developers' ability to supervise and control.
Who should manage safety
AI Benefits and Progress
AI companies can test models, publish risk assessments, add safeguards, and slow development when necessary.
AI Risks and Controls
Amodei said company action alone is insufficient because the risks affect all countries and require government cooperation.
International rules
AI Benefits and Progress
Shared standards could help countries evaluate advanced AI systems and verify their capabilities.
AI Risks and Controls
Governments must also address severe threats through measures such as a ban on using AI to create biological weapons and reporting systems for security incidents.
- Speaker
- Dario Amodei, chief executive of Anthropic
- Main risks
- AI misuse for biological weapons and loss of human control over advanced systems
- Development pace
- Amodei said AI could reach the capability of a 'country of geniuses in a data centre' within one or two years, perhaps less
- Company safeguards
- Anthropic tests models for dangerous capabilities and applies stricter safeguards to its most advanced systems
- Proposed agreement
- Amodei urged countries to agree on specific risks, including banning AI use to create biological weapons
- Global measures
- He proposed common testing standards, capability verification, and a system for reporting AI-related security incidents
- Potential benefits
- Amodei cited AI advances in skilled work and Claude's discovery of a new molecular machine