At an all-hands, Andrew Bosworth says Sam Altman is not happy about Meta getting OpenAI talent and is “being dishonest” about Meta offering $100M bonuses
A Meta exec calls Sam Altman “dishonest,” my conversation with the startup CEO who is suing OpenAI, and AI is coming for games.
The VergeAlex Heath
Context & Ripple Effects
The dispute follows Altman’s public claim that Meta made unusually large offers to OpenAI staff, including alleged $100 million signing bonuses; a subsequent assessment found the specific bonus figure may be overstated while very large total packages remain plausible given Meta’s AI hiring needs scrutiny of the $100 million claim.
Bosworth’s all-hands remarks turn a recruitment contest into a credibility fight between Meta and OpenAI. Altman’s later message to researchers contrasting “missionaries” with “mercenaries” shows how retention is being framed around both mission and financial upside Altman’s mission-versus-money retention appeal.
First-order effects
Meta’s recruitment of OpenAI talent becomes a more visible internal and public pressure point for both companies, while Bosworth directly contests Altman’s account of the offers.
OpenAI researchers face competing narratives about why colleagues leave: Meta’s compensation and opportunity versus OpenAI’s mission and equity upside.
Second-order effects
The public disagreement raises the cost of retaining elite AI staff: OpenAI must reinforce its employee value proposition, while Meta must demonstrate that its offers are credible and strategically motivated rather than simply transactional.
Compensation claims become part of competitive positioning, with outside reporting already separating a disputed signing-bonus figure from the plausibility of larger overall packages the distinction between bonuses and total compensation.
Third-order effects
If this pattern persists, frontier-AI competition will increasingly be shaped by the mobility of a small pool of experienced researchers, making talent retention as strategically important as model development.
The dispute also suggests AI labs will compete for legitimacy as employers: firms may need to pair pay with a persuasive account of mission, ownership, and the work’s long-term impact.
The trend: The AI talent war is evolving from private recruiting into a public contest over compensation, mission, and the credibility of each lab’s strategic story.
In an internal meeting today, a top Meta exec said Sam Altman was “dishonest” for saying that the company is offering $100 million bonuses to poach OpenAI employees. — “we are succeeding at getting talent from OpenAI,” Meta's CTO told staff www.theverge.com/command-line...
i think what meta has to worry about now is basically everyone in AI saying “where's my giant check?” same story in every industry: compensation FOMO folks i've spoken with seem to think this will settle down in a few years after more people dive into “frontier” AI research
internally, meta CTO punching back at Sam Altman and the narrative of Zuckerberg handing out huge checks to any AI researcher with a pulse from what ive heard he's right to say not everyone is getting 100mm offers. that's relatively rare, but exists https://www.theverge.com/... […
this is a high tier psyop from sam. he's talking directly to the market. by floating a number like $100m: - he sets an anchor so absurd that anything meta offers now feels cheap. - he forces meta to either deny or match, both of which hurt them. - he gives his own employees a [im…
De-invest, you say? Are you suggesting, perhaps, that they'd be, say, spending less capital on training llama? Almost like, say, a capital expenditure reduction? Hmm? [embedded post]
“In another extraordinary move, Mr. Zuckerberg and his lieutenants discussed ‘de-investing’ in Meta's A.I. model, Llama, two people familiar with the discussions said.” — A lot of this reads like flailing to me. Particularly the desperate salaries.
Have to wonder if that is a pure admission of defeat or an attempt to get ahead of the danger of becoming a money piñata over their, shall we say, loose governance of training material acquisition [embedded post]
The US needs a highly-capitalized open source AI champion. If Meta stops supporting Llama, pulling the rug out from everyone depending on it, we'll have none. Lame! Though as the cost of frontier training reaches multi-billions, the biz case admittedly does get harder to make...
here is a story inside the last 6 months of Meta's mad dash to save its AI effort after Llama 4 disappointed, any and everything was on the table — including considering de-investing in Llama and using other models from openAI or Anthropic w @CadeMetz https://www.nytimes.com/...
nothing has been decided and meta says they're continuing to support Llama and have multiple releases planned this year. but it explains the degree to which Zuckerberg takes this seriously, working closely with top lieutenants to turn things around as the AI race heats up