Skip to content
Sign in
Technology · Global

Zuckerberg backs Meta’s AI push after Wang’s recommendations

Meta CEO Mark Zuckerberg is backing an AI push informed by Alexandr Wang’s recommendations, building on the company’s strategy of competing with openly available models such as Llama 3.1, released in July 2024.

AI-assisted synthesis · Based on 2 outlets' reporting · Article text read from 1 of 2 outletsHow it's made

View timeline · 9 entries
2 outlets2 regionsLatest reporting 2 min read
Coverage: Too few ratedL 0 · C 1 · R 0 · 1 unrated
File photo
Photo: InvadingInvader / Wikimedia Commons (CC BY-SA 4.0)

What happened

Key facts

  • Zuckerberg said his call with Wang lasted about 44 minutes (per storyboard18.com).
  • Wang said the model achieved state-of-the-art performance on personal and professional agent tasks according to Meta’s internal measurements (per storyboard18.com).
  • Fortune’s headline described Zuckerberg’s AI bet as worth more than $100 billion and said it was about owning everything (per fortune.com, storyboard18.com).

The brief

Alexandr Wang’s recommendations helped reshape Meta’s artificial-intelligence strategy after a call with chief executive Mark Zuckerberg, Storyboard18 reported. The account links Wang’s advice to an aggressive recruitment drive and the development of Meta’s Muse family of models, making the advice one part of the company’s effort to build out its AI work.

Storyboard18 identifies Wang as Scale AI’s cofounder and the leader of Meta Superintelligence Labs. It says he won Zuckerberg’s confidence while Meta was reassessing its strategy following a lukewarm reception for its Llama 4 models. Zuckerberg said the conversation lasted about 44 minutes.

The account also says a WhatsApp message, the phone call and a detailed memo helped set the stage for one of Meta’s biggest artificial-intelligence bets. Fortune’s headline characterizes Zuckerberg’s bet as worth more than $100 billion and frames it as an effort to own everything. That headline does not provide further detail about what the phrase “owning everything” means.

Meta’s Muse work has begun taking shape as a family of AI models, according to Storyboard18. The outlet says the Muse personal AI agent debuted in the United States on September 8. Sensor Tower data cited by Storyboard18 showed the app passed 2.5 million downloads in roughly two weeks and rose to the top of Apple’s free-app rankings.

Wang said the model performed at a state-of-the-art level on personal and professional agent tasks, based on Meta’s internal measurements. This is his characterization of the model, and the account presents the performance assessment as coming from company measurements rather than an independent evaluation.

The account connects the Muse launch and recruitment drive to Wang’s recommendations, but does not detail the hiring numbers or describe the full contents of his memo. It reports no additional statement from Zuckerberg or Wang about the strategy beyond Zuckerberg’s account of the call and Wang’s description of the model’s performance.

Why it matters

  • Meta’s push toward personal AI agents could shift more search, planning, and work tasks into its apps, giving the company new ways to retain users and sell services while putting pressure on rival assistants. If the agents prove useful, users and businesses may rely on Meta’s systems for sensitive tasks, increasing the stakes around accuracy, privacy, and control over their data.

What to watch next

  • Whether Meta develops the Muse family further; Storyboard18 says the team’s work has begun taking shape through those models.
  • Whether Meta provides additional information about the recruitment drive linked by Storyboard18 to Wang’s recommendations.
  • Whether Meta offers performance evidence beyond the internal measurements Wang cited for personal and professional agent tasks.

How it's being covered

Coverage analysisShow details
2 source articles · 2 unique outlets
Outlet political mix: Too few rated· 1 Center · 1 unrated
Where the outlets are based: US/Canada 50% · Asia 50%
Outlet political mix
Too few rated
Left
Center
Right
0 outlets
1 outlet
0 outlets
The marker shows coverage balance: each unique outlet counts once. 1 unrated outlet is not included.
Geography
Where the outlets covering this story are based.
2 unique outletsDominant coverage: US/Canada
Limited coverage
  • Coverage is currently limited to 2 outlets.
How we measure coverage

How outlets' accounts compare

Updat3 does not yet have enough full reports to compare how outlets framed this story.

Timeline

Developments from this story's sources, dated background we compiled and fact-checked, and our earlier coverage. Each row says which.
  1. Mark Zuckerberg and his co-founders launched Facebook.
    Background · compiled and fact-checked by Updat3
  2. Facebook established its Fundamental AI Research (FAIR) group in 2013, with Yann LeCun among its founding leaders.
    Background · from our historical context
  3. The wider generative-AI competition accelerated after OpenAI made ChatGPT publicly available on November 30, 2022.
    Background · from our historical context
  4. In February 2023, Meta announced Llama, a family of large language models, initially distributing it to researchers under an access-based license.
    Background · from our historical context
  5. Meta released Llama 2 with licensing terms that permitted some commercial uses.
    Background · compiled and fact-checked by Updat3
  6. In February 2024, Meta added a “Meta AI” assistant to products including Facebook, Instagram and WhatsApp, connecting its model development to a large consumer-product ecosystem.
    Background · from our historical context
  7. Meta released its Llama 3 language model.
    Background · compiled and fact-checked by Updat3
  8. Meta released Llama 3.1, including a 405-billion-parameter model, for download.
    Background · compiled and fact-checked by Updat3
  9. Meta announced that Alexandr Wang would join its AI team.
    Background · compiled and fact-checked by Updat3

How we got here

Meta’s immediate backdrop was the Llama 3.1 release on July 23, 2024: Meta made its 405-billion-parameter model available for download, extending its strategy of competing through openly available models as well as its own products. Earlier, Meta had released Llama 3 in April 2024 and Llama 2 in July 2023.

That strategy drew on Meta’s long-running investment in artificial intelligence. Facebook established its Fundamental AI Research (FAIR) group in 2013, with Yann LeCun among its founding leaders. In February 2023, Meta announced Llama, a family of large language models, initially distributing it to researchers under an access-based license.

Original sources

1 of 2 articles · Asia · Show all

Every story links to the reporting it was built from. How we choose sources

Spotted a factual error, a broken link or a misleading summary? Tell us.