Dario Amodei Calls for Slower AI Development
Usagevpn.com – Anthropic CEO Dario Amodei calls for AI companies to reduce the pace of their most advanced research, warning that increasingly capable systems require more time for safety work and oversight.
His comments come amid growing debate over “superintelligence”, a theoretical point at which AI could exceed human intellectual capabilities. Amodei argued that progress should remain deliberate enough for developers and society to understand and address potential harms.
“We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain,” Amodei wrote on his personal website.
Finding a balance on advanced AI
Amodei co-founded Anthropic in 2021 with his sister Daniela. The company is known for Claude, its AI model and chatbot. His position is not a demand to stop developing AI, but a call for greater restraint as leading companies build more powerful systems.
He said refusing to develop the technology could prevent people from benefiting from it and could leave advanced AI in the hands of authoritarian governments. At the same time, he said pushing ahead too quickly without adequate safeguards would be reckless.
“Not building the technology deprives humanity of benefits or simply places AI in the hands of authoritarian powers, while building it too fast is reckless. We have sought a middle way,” Amodei said.
Anthropic CEO Dario Amodei calls for more prudence because safety measures must keep pace with models that may become increasingly autonomous and capable of improving their own performance. He said his views had become more cautious in recent months.
Warnings from within the AI industry
The debate follows criticism from AI researcher Jacob Coxon, who left the industry after roughly three years working on advanced-model training. Coxon worked first at OpenAI and joined Anthropic this year, but said neither company was acting responsibly enough.
In posts on X, Coxon criticised competition among major US AI groups to build systems with self-improvement capabilities. He warned that companies were taking serious risks while pursuing increasingly powerful technology.
“Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” Coxon wrote.
Coxon worked on pre-training, the foundational process through which AI models process large volumes of text, images, video and speech recordings. This allows systems to recognise patterns and later generate responses, produce text and images, or answer prompts in chat-based services.
Why calls for a slowdown are growing
AI developers have quickly released models with wider reasoning, coding and tool-use abilities. Critics fear that technical capabilities could advance faster than governments, institutions and the public can establish effective testing, oversight and rules for their use.
Anthropic had already urged a slowdown or suspension of some development efforts in early June. That month, more than 1,000 workers from companies building cutting-edge AI, including Amodei, signed a petition asking the US government to help deliberately set the pace for frontier automated AI development.
Anthropic CEO Dario Amodei calls for a middle course: retaining AI’s possible benefits while allowing more time to reduce the risks associated with highly capable systems.
FAQ: What does a slower AI development pace mean?
Would AI development stop? No. Amodei’s argument is for a more cautious pace, not for ending AI research or denying access to useful applications.
Why do developers want more time? A slower pace could provide additional time for safety testing, oversight and clearer rules before the most powerful systems are widely deployed.
What is superintelligence? It is a theoretical form of AI that could surpass human intellectual abilities, raising concerns about control, safety and the potential consequences of misuse.
