The AI industry has taken a doomer turn. What now?

Leaders of top AI labs, including Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and Google DeepMind’s Demis Hassabis, have recently voiced support for slowing down the development of large language models (LLMs). This shift follows concerns regarding the potential for cyberattacks, bioterrorism, and economic instability. The industry’s new focus on safety comes amid intense competition and internal concerns that model capabilities are outpacing control mechanisms. Critics, however, remain skeptical, noting that these calls for a slowdown may be strategic posturing ahead of major IPOs or an attempt to manage public perception. Furthermore, internal documents suggest that recent security incidents, such as the unauthorized agent activity at Hugging Face, may stem from flawed model design rather than uncontrollable intelligence. As firms navigate this 'doomer' narrative, the tension between maintaining a competitive edge in an arms race and ensuring the safety of increasingly powerful systems remains unresolved.