Sam Altman is back. Improbably and dramatically, the ex-OpenAI CEO returned as CEO late Tuesday. Altman’s counter-coup three board members who sparked his firing and included an agreement to investigate what went down this past weekend.

The new board—which now includes Larry Summers and Bret Taylor—will expand to up to nine members and likely include someone from Microsoft. The A. I.

field will not go back to “normal” after this. OpenAI was already coming into the chaos and will now have to work harder to maintain its lead while facing inspired competition. Though the narrative might frame this as a major win for OpenAI and Microsoft, the reality, as always, is a bit more nuanced.

Here’s how the A. I. field changes after this.

Eighteen thousand Microsoft customers on Azure, and the disintegration of OpenAI would’ve left them scrambling. After Altman’s surprise ouster last week, Microsoft had to restore order to its OpenAI partnership, and it realistically could not have hired OpenAI’s entire staff kept the OpenAI service running. So this is a positive resolution and a relief after a tense few days in which Microsoft played its cards by announcing it had hired Altman and his colleague Greg Brockman to lead a new A.

I division. Now, within OpenAI, there are still some governance issues to resolve. Microsoft doesn’t have an board seat after all this.

But Altman’s return was the least bad option for Satya Nadella & co. , who can now press forward with their industry-leading A. I.

efforts, even if having Altman in-house would’ve paid dividends over time. Companies building on OpenAI technology freaked out this past week. They trusted in a company that almost evaporated in a weekend.

So now, those building on OpenAI are putting contingency plans in place should the situation repeat. The era of model agnosticism is here. Soon, any serious A.

I. company will be able to substitute OpenAI for Anthropic or any other competitor. Startup founders using OpenAI have already told me they’ve started work on it this week.

OpenAI competitors are already trying to exploit the situation. “Utterly insane weekend. So sad.

Wishing everyone involved the very best,” Inflection CEO Mustafa Suleyman this week. In the next breath, the rival A. I.

boss said: “Come run with us!” OpenAI sold the world’s top A. I. researchers on a vision and a safety valve: Join us, help us get closer to human-level artificial intelligence, and if things get unsafe, the board will step in.

It was a win-win proposition that was ultimately a sham. The OpenAI board was poorly structured, almost blew up the company, and the new structure will be less safety-focused. This will open avenues for competitors to recruit researchers who otherwise might’ve gone to OpenAI.

Meta chief A. I. scientist Yann LeCun is already the case that his team’s open-source focus will make it an unlikely winner.

He might be right. Altman’s pushed an A. I.

“safety” agenda in Washington and globally, becoming a lobbying force. OpenAI’s corporate structure lent legitimacy to his efforts. The implicit message: We’re the A.

I. safety company, not the for-profit, please listen to us and consider the following rules. With the nonprofit board’s decision so quickly reversed after pressure from investors (and well-compensated employees), the myth will take a hit.

OpenAI will now become one of the pack, without its special sheen, which will change its ability to influence policy. OpenAI’s board was supposed to save us from an A. I.

apocalypse. Then, it couldn’t think three steps ahead in a boardroom coup. Much of the blame rests with the specific individuals.

But more broadly, it’s hard to imagine anyone will have confidence in our ability to stop harmful A. I. should we develop it.

(And what if the board’s concerns in this area were legitimate?) So the future of the A. I. safety field is in flux.

OpenAI’s chaos may be its own ladder. It moves forward with a board more sympathetic to accelerating A. I.

development. It will work more closely with Microsoft under the new structure, with fewer speed bumps along the way. And it may have en route.

But the chaos will also be a ladder to those OpenAI was once lapping. And some competitor—whether it’s Anthropic, Inflection, Google, or others—will inevitably exploit the moment and.