Recommended Free Tools
In the final week of June 2025, Meta recruited at least eight OpenAI researchers as Mark Zuckerberg assembled a new superintelligence organization. The count came from several reports published at different points, so it is best understood as a burst of at least eight reported hires—not a single official roster or a new August 2026 event.
What happened
The sequence began with Meta hiring Trapit Bansal, an OpenAI researcher associated with reasoning-model work, according to TechCrunch. Bloomberg then reported four more hires—Jiahui Yu, Shuchao Bi, Shengjia Zhao and Hongyu Ren—on June 28: Bloomberg. Coverage subsequently identified Zurich-based researchers Lucas Beyer, Alexander Kolesnikov and Xiaohua Zhai, bringing the reported total to at least eight: Business Standard.
OpenAI research chief Mark Chen reportedly told employees that the company was reviewing compensation and recruiting practices after the departures. Meta then placed the hires inside a broader Meta Superintelligence Labs effort rather than treating them as an isolated team acquisition.
Who the eight researchers were
| Person | Previous role or association | Reported significance | How the hire was reported |
|---|---|---|---|
| Trapit Bansal | OpenAI researcher | Associated with reasoning-model work | TechCrunch report |
| Lucas Beyer | OpenAI, Zurich | Researcher in Meta’s reported Zurich group | Business Standard/WSJ-based coverage |
| Alexander Kolesnikov | OpenAI, Zurich | Researcher in the same reported cluster | Business Standard/WSJ-based coverage |
| Xiaohua Zhai | OpenAI, Zurich | Researcher in the same reported cluster | Business Standard/WSJ-based coverage |
| Jiahui Yu | OpenAI researcher | Associated with multimodal and computer-vision work | Bloomberg and related coverage |
| Shuchao Bi | OpenAI researcher | Contributor associated with GPT-4o voice mode and o4-mini | Named in Zuckerberg’s hiring memo as reported by WIRED |
| Shengjia Zhao | OpenAI researcher | Later named chief scientist of Meta Superintelligence Labs | Bloomberg and Meta reporting |
| Hongyu Ren | OpenAI researcher | Recruited into Meta’s new frontier-model group | Bloomberg and related coverage |
These were researchers with different specialties and seniority. The evidence does not support describing all eight as founders, executives or creators of OpenAI’s flagship systems.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
The reported timeline
- June 17, 2025: Sam Altman said Meta had attempted to recruit OpenAI employees with offers reportedly reaching $100 million, while saying none of OpenAI’s best people had accepted at that point. TechCrunch.
- June 26: TechCrunch reported Bansal’s move to Meta’s reasoning-model effort.
- June 28: Bloomberg reported the four hires Yu, Bi, Zhao and Ren.
- June 29–30: Reports counted at least eight hires, including the three Zurich researchers. WinBuzzer.
- June 30–July 2: Zuckerberg announced Meta Superintelligence Labs and a wider group recruited from OpenAI, Google, Anthropic and elsewhere. Fortune.
- July 25: Meta identified Zhao as chief scientist of the new group. TechCrunch.
- July 30: Meta publicly described its “personal superintelligence for everyone” strategy. Meta.
Why Meta paid so aggressively
Closing a perceived capability gap
Meta had substantial AI resources but was competing with frontier labs such as OpenAI, Google and Anthropic. Hiring people who had worked directly on advanced models offered a faster route to expertise in training, evaluation, scaling and inference than building every capability internally.
Building a dense team
A cluster of researchers who already know one another can start collaborating more quickly and can make additional recruiting easier. The hires also supplied credibility for a new organization whose mission was to pursue frontier models and, in Meta’s words, personal superintelligence.
Combining research with distribution
Meta’s potential advantage was its consumer reach and products, not just laboratory research. The company presented Superintelligence Labs as a structure spanning foundation models, FAIR, products and new frontier-model work. Meta also recruited Scale AI chief executive Alexandr Wang after investing in Scale AI, and brought in other prominent technology executives and researchers. AP.
What the compensation reports actually said
WIRED reported that selected top candidates were offered packages worth as much as $300 million over four years, with more than $100 million in first-year total compensation in some cases. Those figures describe reported maximum packages, not a standard rate paid to every recruit. They may include salary, equity, bonuses and other elements; they should not be described as a universal $100 million signing bonus.
Altman’s earlier public comments about nine-figure offers came before the later reported hires. Meta CTO Andrew Bosworth reportedly told employees that the structure was more complicated than a simple signing payment. No public evidence establishes what each of the eight researchers received, how much was guaranteed, or how much had vested.
How OpenAI responded
OpenAI’s reported internal response focused on retention and compensation. Chen’s message described a review of pay structures after the departures, while Altman publicly criticized Meta’s recruiting campaign. The Information characterized Altman’s comments as calling the effort distasteful.
Those reactions show that OpenAI considered the campaign a serious organizational threat. They do not prove that OpenAI’s models were damaged, that the company faced existential danger, or that the departing researchers transferred proprietary technology.
Talent is not the same as technology
Researchers can bring general expertise, professional relationships and experience with frontier-model practices. They cannot automatically bring an employer’s confidential code, model weights, training data, infrastructure or trade secrets. Confidentiality, garden-leave and non-solicitation terms could raise legal and practical questions, depending on the employee and jurisdiction.
That distinction matters financially and strategically: a high-priced hiring burst can shorten the time needed to form a team, but it does not purchase a functioning research organization, proven infrastructure or product-market fit.
The broader AI talent war
Meta’s campaign was one episode in a wider escalation. The company also recruited Nat Friedman and Daniel Gross, pursued talent linked to Safe Superintelligence, and competed with Google, Anthropic, Microsoft, Apple, OpenAI and startups such as Perplexity and Thinking Machines Lab. CNBC reported on Meta’s efforts involving Safe Superintelligence.
The pattern was broader than ordinary hiring: extreme compensation, chief-executive involvement, investments tied to recruiting and so-called reverse-acquihire tactics. It also raised labor-market concerns because a small number of frontier labs could bid for scarce specialists at levels far beyond most AI employers.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why the raid did not settle the competition
In August 2025, WIRED reported that three recent Meta superintelligence hires had left, with two returning to OpenAI: WIRED. The departures do not erase Meta’s achievement in recruiting the group, but they undermine any claim that the hiring count alone proved Meta had won the talent war.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteRetention depends on autonomy, research scope, reporting lines, infrastructure, mission and culture, as well as the portion of compensation that is actually guaranteed or vested. A team assembled quickly can also face friction when folded into a large company’s existing organization.
Bottom line
Meta’s June 2025 recruitment of at least eight OpenAI researchers demonstrated how valuable frontier-AI expertise had become and how aggressively Zuckerberg was willing to compete for it. The hires strengthened Meta’s new superintelligence organization, but they did not transfer OpenAI’s models or guarantee a better system. Later departures showed that money and prestige can open a recruiting campaign without guaranteeing durable advantage.
Frequently Asked Questions
Did Meta hire exactly eight OpenAI researchers?
Reports identified at least eight during a roughly one-week burst in late June 2025. Because the count combined announcements, internal references and unnamed-source reporting, “at least eight” is more accurate than an exact final total.
Did every researcher receive a $100 million or $300 million offer?
No. WIRED reported packages as high as $300 million over four years for selected candidates, with more than $100 million in first-year total compensation in some cases. Public reporting does not establish a uniform package for the eight hires.
Did the researchers bring OpenAI’s technology to Meta?
There is no evidence that they transferred OpenAI code, model weights, datasets or trade secrets. Employees may use general knowledge and experience, subject to confidentiality and other obligations.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
