Anthropic and OpenAI CEOs Call for Slowing AI Development Amid Safety Concerns

Anthropic CEO Dario Amodei stirred the AI community over the weekend when he penned a detailed essay advocating for a slower pace of development in advanced AI models. Rival OpenAI CEO Sam Altman echoed Amodei’s sentiments in a subsequent post on X, stating his agreement.

Amodei’s call to action follows warnings from two AI researchers who claim that the technology could pose a grave threat. On Tuesday, Jacob Coxon, a former Anthropic worker, resigned from the company, citing safety concerns. In his X post, Coxon contended that both Anthropic and OpenAI are racing to develop “self improving superintelligence” without sufficient oversight.

Hubinger, Anthropic’s Alignment Science Lead, also chimed in, agreeing with Coxon that there is a greater than 10% chance that the technology could “kill all humans.” Coxon’s and Hubinger’s comments have raised alarms across Silicon Valley and even in Washington, DC.

However, former President Donald Trump pushed back on the safety concerns, stating in a Sunday tweet that the United States is “leading China in AI” and that he wants to “keep it that way.”

The AI warnings echo recent revelations of hacking attempts by Anthropic, OpenAI, and Meta. Amodei’s essay underscores the potential for recursive self improvement, or AI that can develop better AI models on its own, which could outpace human understanding and control.

Amodei argues that frontier AI companies should commit to embedding third party evaluators to assess AI models and ensure adherence to safety practices. He also calls for safety standards in democratic countries and a coordinated approach with authoritarian governments to address AI safety.

Amodei’s essay comes as Anthropic prepares for an initial public offering later this fall, amid a fierce race among leading AI labs to develop the most advanced models.

This development highlights the growing concerns about the rapid advancement of AI and the urgent need for better safety measures in the industry.

Source: https://finance.yahoo.com/technology/article/anthropic-openai-ceos-call-for-pacing-ai-development-after-researcher-says-technology-could-kill-all-humans-163235871.html