Archive · 13 Sep 2026
Anthropic CEO Urges Slower AI Development Over Human Safety Risks
Dario Amodei leads a company called Anthropic that builds artificial intelligence.
He says powerful AI could become dangerous if people develop it too quickly.
One former researcher, Jacob Coxon, warned that AI might seriously threaten humans within the next decade.
Amodei said he agrees with many of Coxon’s concerns.
He believes the future is not decided by one fixed probability.
Instead, people can choose safer or riskier paths.
He wants companies and governments to work together and slow down enough to add safety protections.
OpenAI leader Sam Altman supported slowing the development of the most advanced AI systems.
Elon Musk also said Amodei was right.
Anthropic CEO Dario Amodei said AI could pose serious risks, including potentially threatening humanity.
Amodei argued that governments and companies should slow AI development to create stronger safety measures.
Former Anthropic researcher Jacob Coxon warned that advanced AI could kill humans within the next decade.
Amodei said outcomes depend on choices made by the AI industry, companies, and governments.
OpenAI CEO Sam Altman and Elon Musk supported caution, while OpenAI agreed to use independent evaluators.
- Who
- Anthropic CEO Dario Amodei, former researcher Jacob Coxon, OpenAI CEO Sam Altman, Elon Musk, and other AI industry and government leaders.
- What
- They debated whether advanced AI could threaten humans and whether development should be slowed for safety.
- When
- Amodei discussed the issue in an interview on Saturday; related warnings and statements were made recently.
- Where
- In an exclusive interview with CNN's Anderson Cooper and in public statements on X and company communications.
- Why
- The discussion concerns risks from increasingly powerful AI, including loss of control, cyberattacks, bioterrorism, and economic disruption.
Slower, Coordinated Development
Continued Rapid Development
Pace of AI progress
Slower, Coordinated Development
Amodei argues that the industry should slow capability improvements to gain time for safety measures and cooperation.
Continued Rapid Development
The article says competition among companies is driving rapid progress, and Amodei warned that not all companies may agree to slowing down.
How serious are the risks?
Slower, Coordinated Development
Coxon and Amodei warn that advanced AI could threaten humanity and create risks involving loss of control, cyberattacks, bioterrorism, and economic disruption.
Continued Rapid Development
Amodei said the probability of something going wrong is not necessarily high and depends on which choices companies and governments make.
Industry responsibility
Slower, Coordinated Development
Amodei and Altman support cooperation, including independent evaluators with employee-like access to advanced AI systems.
Continued Rapid Development
Amodei said a race driven by commercial incentives could worsen risks, while acknowledging that some companies may not support a coordinated approach.
- Main speaker
- Dario Amodei, CEO of Anthropic
- Former employee raising concerns
- Jacob Coxon, who previously worked at Anthropic and OpenAI
- Potential AI risks
- Loss of control, cyberattacks, bioterrorism, and serious economic disruption
- Amodei's proposal
- Slow the pace of frontier AI capability improvements
- OpenAI's response
- Sam Altman said OpenAI would use independent evaluators with employee-like access
- Other support
- Elon Musk wrote that Amodei was right
- International warning
- United Nations human rights chief Volker Turk called for strong AI safety guarantees