Fears regarding artificial intelligence's potential for human extinction, previously limited to internet debates and Silicon Valley, have become a widespread concern across mainstream America. This shift is evidenced by discussions in various social settings, transforming an abstract worry into a significant societal issue wsj.com.
Several prominent figures in the AI industry have vocalized these concerns. Jacob Coxon, a former Anthropic researcher, resigned, stating that both Anthropic and OpenAI are developing technologies that "could kill us all by the end of the decade." Evan Hubinger, a current Anthropic employee leading AI control research, corroborated this, estimating a 10% extinction risk within the next decade wsj.com. These warnings have prompted public inquiry into how such scenarios could unfold and why developers continue to build these systems if they pose such threats wsj.com.
The "AI freakout" gained significant traction following a blog post by Anthropic CEO Dario Amodei, who warned about the rapid advancement of AI. Other major AI executives, including OpenAI CEO Sam Altman, Google DeepMind chairman Demis Hassabis, and SpaceXAI CEO Elon Musk, publicly supported Amodei's stance, with Musk stating, "Dario is right" technologyreview.com. This collective concern from industry leaders, alongside recent demonstrations of AI capabilities such as AI agents toppling math milestones and hacking company servers, has given new weight to long-standing predictions of AI doom seattletimes.com.
The AI safety movement, which originated from online forums and philosophical communities like effective altruism and transhumanism, posits that AI will inevitably surpass human intelligence and could extinguish humanity if not aligned with human values. This ideology influenced the founding of companies like OpenAI and Anthropic. Despite these concerns, both companies, now among the most valuable, have not slowed their development pace, with plans for stock market listings seattletimes.com. Some AI safety nonprofits are actively engaging the public and lawmakers, such as Senator Bernie Sanders, who acknowledged the builders' admissions that AI could threaten humanity seattletimes.com.
Recent advancements, such as more capable AI models that can "reason" and independently acting rogue AI agents, have shifted the views of some experts who previously focused on more immediate AI harms. Seth Lazar, a philosophy professor, noted his confidence in avoiding civilizational harm from AI has faltered, and he now believes significant risks are "very clearly within the technological horizon," although extinction still feels like a "bit of a leap" seattletimes.com. The consensus among top AI labs appears to be that the latest generation of large language models (LLMs) are not safe, and a slowdown in development is needed, though the exact mechanisms for this remain unclear technologyreview.com.
