Anthropic CEO Says AI Industry Understated Risks

Anthropic CEO and co-founder Dario Amodei says the artificial intelligence industry has for too long understated the risks associated with increasingly powerful AI systems.

Anthropic CEO and co-founder Dario Amodei has called for a slower pace of AI development and stronger safeguards as the technology becomes increasingly capable. [Photo: TechCrunch Flickr]

“I won’t lie to you — there are real dangers,” Amodei told CBS News correspondent Jo Ling Kent. “And I think for too long the industry lied to people about the fact that this technology had risks.”

One of the most serious concerns is the possibility that highly advanced AI could eventually pose an existential threat to humanity.

Former Anthropic researcher Jacob Coxon warned on social media that Anthropic and OpenAI are “gambling with our lives” by developing self-improving “superintelligence.”

Coxon told CBS News that such systems could become capable of causing catastrophic harm, comparing the prospect to scenarios depicted in science-fiction films.

His comments represent his personal assessment, rather than an established prediction about the future of AI.

Amodei said he had not fully anticipated how quickly AI capabilities would advance.

“I don’t think I fully just appreciated what it would actually be like when the progress was as fast as it was,” he said during the CBS News interview at Anthropic’s headquarters in San Francisco.

Amodei described AI development as following an exponential curve that is becoming steeper.

He stressed that the situation does not call for immediate panic or a complete halt to development, but said the pace should be slowed.

“It doesn’t mean we need to panic today. It doesn’t mean we need to shut it all down. But I would say it’s a warning sign,” Amodei said. “It’s a warning sign that we need to slow down.”

In an essay published over the weekend, Amodei called for the technology industry to work together to slow the pace of AI capability development.

He also said Anthropic would give independent evaluators permanent, employee-like access to its models to help verify whether the company is following its stated safety practices.

Amodei supports federal regulation but rejected a complete ban on advanced AI, arguing that the technology could bring major benefits and that development would continue in other countries.

He said international cooperation, potentially modeled partly on nuclear-arms agreements, could eventually be needed to manage the risks.