Anthropic CEO Dario says AI companies have been telling lies, reveals real dangers of AI

In the past week, the debate around AI has quickly shifted from the most powerful AI models to AI potentially killing humans in the next few years. Anthropic CEO Dario Amodei has now claimed that AI companies had lied about risks associated with this technology for years. He accepted that there were some real dangers that may come with AI.

In an interview with CBS, Dario Amodei addressed fears surrounding advanced AI models, and the fact that there were risks that could not be ignored. “I won't lie to you, there are real dangers. And I think for too long the industry lied to people about the fact that this technology had risk,” he said.advertisement❮❯ Read Full StoryEnd of humanity is biggest risk of AI, says AmodeiAs per the Anthropic CEO, one of the major risks with AI may be that it ends humanity entirely. “The biggest risk could be the end of humanity,” Amodei added.

Dario Amodei’s comments come after former Anthropic researcher Jacob Coxon claimed on X that companies like Anthropic and OpenAI were gambling with our lives. “The people building AI earnestly believe that it could kill us all by the end of the decade,” he added.

Responding to Coxon's warning, Amodei said he had not "fully just appreciated what it would actually be like when the progress was as fast as it was.” The Anthropic chief clarified, however, that the current pace of progress was more of a warning, and not something that should make people panic. “It's a warning sign. It's a warning sign that we need to slow down,” he said.

In a separate interview with CNN, Dario Amodei said that he agreed with Coxon "much more than" he disagreed, though he did not say a disastrous outcome was certain. Coxon is not the only one who has expressed fears of AI ending humanity. OpenAI safety researcher Marcus Williams recently claimed that there may be a 70 per cent chance AI will make humans extinct within the next three years.

Amodei has also rejected reducing the risk to a simple percentage, saying the outcome depends on the path the industry takes. "If we build in the right way, I think the probability of something bad happening is very low," he said. "If we build in the wrong way, the probability of something bad happening is very high." His call to pace AI development has since drawn support from other industry figures, with OpenAI chief executive Sam Altman saying his company would also commit to independent evaluators with employee-like access, while Elon Musk wrote on X: "Dario is right."Anthropic won’t stop releasing new AI models

You may wonder whether such warnings would lead to AI companies like Anthropic to potentially stop releasing new models. However, Dario Amodei believes that should not be the case. “What it means is that we need to make sure that every generation of models that we release is properly tested,” he said.

In an essay shared on X, Amodei said that Anthropic was going to “provide third-party evaluators with permanent, employee-level access” to its systems so they can assess model behaviour, training processes and whether the company is following the safety practices it has committed to.

Though Amodei is not the only one suggesting ways to check on AI development. Among the proposals under discussion are a ban on superintelligence and a mandatory AI "kill switch" that could shut down risky systems. Amodei said a kill switch "could be a good idea", but added that he did not support a complete ban. “One, the idea that we would never get the incredible benefits, would never get the medical cures, that doesn't sit right with me. Second, this technology will be built in other countries, including authoritarian countries,” he explained.

When asked whether the United States should still slow down if China did not, he replied: "This is the toughest dilemma. This is the toughest aspect of our situation."

On the question of who should control the technology, Amodei said it had always seemed strange to him that AI was being built by a private company, and added that he was uncomfortable with that reality. The Anthropic CEO claimed that he would be open to oversight by "the right combination of governments,” while warning that a single government could abuse the technology as easily as a single company could.- EndsPublished By: Armaan AgarwalPublished On: Sep 14, 2026 09:20 ISTAlso Read | Anthropic CEO urges AI companies to slow model development amid fears over misuseAlso Read | No longer merely aiding humans, but on course to become them: Ex-Anthropic researcherAlso Read | Ex-White House AI czar tells Anthropic, OpenAI to stop pretending and slow AI