Elon Musk thinks the world is woefully unprepared for the impact of artificial intelligence. On Sunday, he agreed that the technology will hit people “like an asteroid,” and he revealed that he used his only one-on-one meeting with then President Barack Obama to push for A.I. regulation.

The Twitter and Tesla CEO made the comments in response to a tweet from A.I. software developer Mckay Wrigley, who wrote on Saturday: “It blows my mind that people can’t apply exponential growth to the capabilities of AI. You would’ve been called a *lunatic* a year ago if you said we’d have GPT-4 level AI right now. Now think another year. 5yrs? 10yrs? It’s going to hit them like an asteroid.”

Musk responded: “I saw it happening from well before GPT-1, which is why I tried to warn the public for years. The only one on one meeting I ever had with Obama as President I used not to promote Tesla or SpaceX, but to encourage AI regulation.” Obama had dinner with Musk in February 2015 in San Francisco.

This week, Musk responded to news about Senate Majority Leader Chuck Schumer laying the groundwork for Congress to regulate artificial intelligence.

“Good news! AI regulation will be far more important than it may seem today,” Musk tweeted.

According to the Financial Times, Musk is developing plans to launch an A.I. startup, dubbed X.AI, to compete against Microsoft-backed OpenAI, which makes generative A.I. tools, including the A.I. chatbots ChatGPT and GPT-4 and the image generator DALL-E 2.

Musk is also reportedly working on an A.I. project at Twitter.

A few weeks ago, Musk called for a six-month pause on developing A.I. tools more advanced than GPT-4, the successor to ChatGPT. He was joined in signing an open letter by hundreds of technology experts, among them Apple cofounder Steve Wozniak. The letter warned of mass-scale misinformation and the mass automation of jobs.

The power of A.I. systems to automate some white-collar jobs is in little doubt. A Wharton professor recently ran an experiment to see what A.I. tools could accomplish on a business project in 30 minutes and called the results “superhuman.” Meanwhile some remote workers are apparently taking advantage of productivity-enhancing A.I. tools to hold multiple full-time jobs, with their employers none the wiser. But fears that in the long run A.I. will replace many jobs are mounting.

Musk cofounded OpenAI in 2015 as a nonprofit, but he parted ways with it after a power struggle with CEO Sam Altman over its control and direction, according to the Wall Street Journal.

He tweeted on Feb. 17 that OpenAI was created as an open-source nonprofit “to serve as a counterweight to Google, but now it has become a closed source, maximum-profit company effectively controlled by Microsoft. Not what I intended at all.”

Altman himself has warned frequently about the dangers of artificial intelligence. Last month in an ABC interview, he said that other A.I. developers working on ChatGPT-like tools won’t put on the kind of safety limits his firm has—and the clock is ticking.

Musk has long believed oversight for artificial intelligence is necessary, having described the technology as “potentially more dangerous than nukes.”

“We need some kind of, like, regulatory authority or something overseeing A.I. development,” he told Tesla investors last month. “Make sure it’s operating in the public interest.”