Anthropic CEO Dario Amodei has shared his views on both the potential risks and benefits of artificial intelligence, noting a 25 percent chance that AI could have catastrophic consequences, while there is a 75 percent probability that it could produce highly positive outcomes. He made these remarks at the Axios AI + DC Summit.
Amodei clarified that when he referred to AI going “really, really badly,” he was not talking about minor glitches like autocorrect errors. Instead, he meant scenarios that could destabilize societal systems, create existential risks, or lead to severe misuse of AI technology. He has also previously warned that AI might replace up to half of all entry-level white-collar jobs and expressed concerns about exporting advanced chips to China, highlighting the economic and geopolitical stakes involved in AI development.
By presenting the issue in probabilistic terms, Amodei emphasized that while AI carries inherent risks, human decisions and governance can shape outcomes. This approach provides a balanced perspective, acknowledging both the dangers and the opportunities that AI presents.
Separately, it was reported earlier this month that Anthropic PBC has agreed to a minimum payout of $1.5 billion plus interest to settle a class-action lawsuit. The case involved authors who accused the AI company of illegally scraping millions of copyrighted books to train its language models.
The settlement, filed in a San Francisco federal court, is among the largest to date involving AI and intellectual property. A trial had been scheduled for December, with potential damages that could have reached trillions of dollars, posing a serious risk to the company’s future.
The agreement covers claims from up to seven million authors, with initial payouts expected to be around $3,000 per book for an initial set of 500,000 works. Additional claims could increase total compensation. Anthropic has also committed to deleting any disputed data that was allegedly obtained without permission. Legal experts described the settlement as unprecedented. Justin Nelson of Susman Godfrey stated that the deal sends a clear message to AI companies and content creators that using copyrighted material without consent is unacceptable.
Although the financial impact is significant, analysts noted that Anthropic may have avoided a far worse outcome. Tod Cohen, partner at Manatt, Phelps & Phillips, observed that the settlement represents less than one percent of the company’s valuation. He added that it appears to be a favorable result for Anthropic, as it avoids dismantling its models and only requires the removal of the contested data.