AI safety experts warn of serious risks to humanity

A senior researcher at Anthropic, one of the world's leading AI companies, has warned that there is more than a 10 percent chance that artificial intelligence could pose an existential threat to humanity by the end of this decade. This warning came just hours after Jacob Coxon, a researcher who has worked on training AI systems at Anthropic, publicly resigned from the company.

In his resignation announcement, Coxon expressed deep concern that Anthropic and its competitors are moving too quickly to build "superhuman" AI systems—machines that would surpass human intelligence—without adequate safety measures in place. He accused the companies of "racing straight to self-improving superintelligence and gambling with our lives." By self-improving, he means AI systems that can continuously enhance their own capabilities without human supervision, which experts worry could spiral beyond human control.

This dispute highlights a growing tension in the AI industry. While companies compete to build more powerful systems faster, some of their own researchers are becoming increasingly worried about the risks. Coxon's departure suggests that concerns about safety are not just theoretical debates—they're serious enough that talented experts are willing to leave their jobs over them.

The underlying question is urgent: how fast should companies develop AI, and how much should safety checks slow down that progress? For ordinary people, this matters because the decisions being made now in AI labs could shape what AI becomes in our everyday lives—and whether it remains under human control.