Washington: Leading voices in the artificial intelligence sector continue to warn that advanced AI could pose a severe risk to humanity if development goes in the wrong direction.
Responding to the debate, Anthropic CEO Dario Amodei said the danger is not fixed but depends on choices made by companies and governments. He argued that if the industry stays on the “right path,” the chance of catastrophe remains low, but if it takes the “wrong path,” the risk could rise sharply.
Amodei also addressed comments by Jacob Koxon, a former Anthropic employee, who posted online that some AI builders genuinely believe AI could destroy humans by 2030. Amodei said he largely agrees with Koxon’s concern, adding that the criticism was aimed at the overall speed of the AI sector rather than at Anthropic specifically.
He stressed that even slowing down carries risks if the technology ends up in the hands of “wrong actors,” and called for coordinated action by the AI industry and countries worldwide.
Amodei outlined three safety priorities: ensuring safety before accelerating AI development, continuous independent evaluation of AI systems, and global cooperation to create shared rules and standards. OpenAI founder Sam Altman said he agrees on controlling the pace of frontier AI and noted internal discussions, including appointing independent evaluators. Elon Musk also publicly supported Amodei’s view.
