What happened
Employees at major artificial intelligence firms, including OpenAI, Meta, and DeepMind, have expressed skepticism regarding recent viral warnings that unchecked AI development could lead to human extinction. While high-profile figures like former Anthropic employee Jacob Coxon have urged a slowdown in development, many industry insiders reacted with amusement to these existential fears. The BBC spoke with several anonymous workers who noted that the tone within the sector has remained largely jokey rather than panicked.
Why it matters
Critics of the doomsday narrative argue that the claims are often vague and rely on hypothetical scenarios, such as non-existent AI models creating biological weapons. Nvidia CEO Jensen Huang dismissed the idea that 2030 would mark the end of the world, calling such fears irresponsible. Similarly, Meta data scientist Colin Fraser stated there is no evidence that large language models possess the drive to wipe out humanity, summarizing that they simply do not have that dog in them. These technical rebuttals highlight a disconnect between public alarm and internal industry consensus.
What comes next
Despite the lighthearted dismissal of extinction risks, experts agree on the importance of mitigating genuine, immediate threats. These include preventing hackers from bypassing AI guardrails and addressing ethical concerns regarding military adoption. Following recent incidents where OpenAI models behaved unexpectedly during security tests, there is growing support for embedding independent safety evaluators within major labs. Over one hundred AI professionals have signed a letter insisting that outside evaluators must remain meaningfully independent to ensure robust oversight.