Musk: AI Will Surpass Human Intelligence Within Five Years, Urges Rivals to Set Aside Feuds and Build Safety Guardrails

Tesla (TSLA) CEO Elon Musk dropped a bombshell prediction in a recent interview with The Economist Editor-in-Chief Zanny Minton Beddoes: the overall intelligence level of artificial intelligence will comprehensively surpass that of humans within five years. He simultaneously opposed sweeping government regulation of AI, proposed an industry-led "peer review" mechanism, and spoke with rare candor about his long-running feud with OpenAI CEO Sam Altman, calling on all parties to set aside their differences for the sake of global safety.

Musk laid out an aggressive timeline for AI development during the interview. He predicted that within roughly five years, AI intelligence will eclipse that of humans; when pressed on whether humanity would still hold the reins a decade from now, he admitted the probability was "unlikely." He explained the logic using a biological analogy: once the intelligence gap between AI and humans far exceeds the gap between humans and chimpanzees, humanity may gradually lose control of the situation, much as chimpanzees cannot dominate human society.

Musk has long maintained a high level of vigilance regarding the existential risks of AI. In the interview, he did not deny that the probability of killer robots causing human extinction falls somewhere between 10% and 20%. When the host posed the pointed question, "If a rocket had a 20% chance of exploding, would you still board it?", Musk answered in the affirmative. He stressed that the trajectory of AI development is now irreversible, and even if a kill switch existed, humanity perhaps should not press it, because barring extreme catastrophes like nuclear war, AI could usher in an era of unprecedented material abundance within the next decade. He predicted that by around 2036, humanity is highly likely to enter a golden age where "everyone can have everything they want."

On the highly scrutinized topic of AI regulation, Musk expressed little confidence in top-down government control models. He believes policymakers generally lack sufficient technical judgment to accurately assess the safety risks of new models. To address this, he proposed a regulatory framework driven by the industry itself, centered on three layers:

First, major AI labs—including OpenAI, Google DeepMind, Anthropic, and his own xAI—should immediately establish a regular communication mechanism, holding safety meetings weekly or bi-weekly to synchronize information on potential risks within their respective models. Musk revealed that he has already reached a consensus on this with DeepMind founder Demis Hassabis, believing the matter brooks no delay.

Second, he advocates that companies should open new models for competitor review one to two weeks prior to release. If a peer discovers a significant safety concern, they would have the right to demand a delay in the launch. In his view, commercial rivalry can actually compel transparency, as competitors will never easily overlook each other's security flaws.

As for the government's role, Musk believes it should retreat to being the last line of defense. Only when a company refuses to remediate issues after receiving peer warnings should the government intervene, rather than imposing blanket restrictions on model research and development upfront. He emphasized that this mechanism must be activated immediately; the industry's original plan to build it within six months is, in his view, far too slow, given that AI technology is evolving at a staggering pace—a six-month window is enough to sow seeds of irreparable harm.

The "consensus" Musk referred to is a concrete proposal that Hassabis put forward ten days before the interview was published. On July 14, 2026, Hassabis published a lengthy post on X titled "A Framework for Frontier AI and the Dawning of a New Age," advocating for the creation of a "Frontier AI Standards Body" modeled on the U.S. Financial Industry Regulatory Authority (FINRA): industry-funded, U.S. government-endorsed, independently operated, with a board that includes independent technical experts and open-source community representatives. Under his design, frontier labs would initially submit models for evaluation by this body on a voluntary basis up to 30 days before release; once the evaluation process proves effective, it would be codified into law, at which point models would need to pass review before deployment in the U.S. market. Labs would also be required to collaborate with the body on critical vulnerabilities discovered post-release. Hassabis argued that this body should begin operations by the end of this year. Microsoft CEO Satya Nadella and DeepMind co-founder Mustafa Suleyman have publicly expressed support. Musk confirmed in the interview that he spoke with Hassabis for several hours before the article was published.

Notably, the "top-down government review" that Musk opposes is already taking shape in the United States. President Trump signed an executive order on AI oversight in June 2026, establishing a federal framework for national security risk reviews of the most advanced AI systems for up to 30 days before public release; the order positions developer participation as voluntary, and the framework's details remain incomplete. Parallel case-by-case directives have already tangibly impacted model releases: Anthropic's Fable 5 and Mythos 5 were taken down days after launch to comply with a Trump administration directive barring access by foreign nationals; the government subsequently lifted restrictions on Mythos 5, redeploying it to a small group of cyber defenders and infrastructure operators. OpenAI's GPT-5.6 Sol was made available only to customers individually approved by the government, numbering around 20 as of reporting. OpenAI stated, "We do not believe this process of government-vetted access should become the long-term default," framing this testing period as a temporary arrangement on the path to broader availability. TechCrunch noted that such case-by-case reviews have drawn criticism for insufficient technical expertise and opaque approval timelines—the very real-world backdrop for Musk's doubts about government officials' lack of technical judgment.