Anthropic exit sparks AI fear aimed at wrong risks

The resignation of Jacob Coxon from Anthropic this week did not happen in a vacuum. Back in March 2024, the author wrote about how the term "AI safety" was expanding to include issues beyond "extinction." At that time, advocate Eliezer Yudkowsky told the author that using the term was a bad idea. The article now says it still is.

That earlier argument explains the current complaint: Coxon's resignation is being used as a catch-all piece of ammunition for anyone who has anything against AI, pulling in data centers' environmental costs, hacking, and AI's impact on the job market. The article notes that many AI safety advocates would actually prefer data centers be built in our own backyards, since moving them offshore makes them harder to monitor and regulate — the problem is energy waste, not their existence. It also warns of a cybersecurity crisis the world is failing to practically address, and says that while AI will likely create many new jobs, little has been done to ensure everyday people outside tech benefit economically.

The stakes come down to whether lawmakers, regulators and tech-company leaders can tackle each problem with sober precision. If they don't, the industry risks having to follow rules that don't solve anything, or worse, slowing down innovation — including on developments that could make AI safer and more beneficial to the economy. That outcome appears to hinge on whether the debate can be separated into distinct issues rather than treated as one."}