Meta is widely regarded as the world’s largest social media company, yet it’s struggling to retain top talent in its AI division. The company has lost 11 of the 14 original authors credited in the whitepaper that introduced its AI model Llama in 2023, as reported by Business Insider.
Getting to know the team behind Llama
Meta officially introduced its Llama AI model in February 2023 in a 27-page whitepaper outlining the model’s technical foundation. The Meta paper credited 14 researchers for their contributions to the AI model’s development, including:
- Naman Goyal: Worked with Meta for 6+ years before joining Thinking Machines Lab, a company focused on developing modular AI architectures for specific applications.
- Eric Hambro: Worked with Meta for 3+ years before joining Anthropic, the team behind the Claude family of LLMs.
- Marie-Anne Lauchaux: Worked with Meta for 5 years before becoming a founding member of Mistral, a company focused on natural language processing, coding, and text generation.
- Aurélien Rodriguez: Worked with Meta for nearly 3 years before joining Cohere, a company specializing in generative and retrieval AI.
All four — along with several other contributions — have since left Meta to work with competing AI companies. According to Business Insider, only three of the original 14 authors remain at Meta: Hugo Touvron, Xavier Martinet, and Faisal Azhar.
SEE: Meta, Google, Amazon, Microsoft Staff Reveal How Tech’s Dream Jobs Turned Cutthroat
More must-read AI coverage
- SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?
- SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?
- Why Data, Not Models, Determines AI Success
- The Rise of the AI-Native Factory: How Physical AI Is Transforming Manufacturing
Exploring potential legal and ethical issues
Meta’s AI ambitions are further complicated by a growing number of lawsuits alleging the use of copyright-protected materials in the training of its LLMs. While Meta has denied the claims, critics continue to question the transparency of its data practices.
According to the original Llama paper: “We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively.”
Meta is also under fire for allegedly sourcing the content from illegal, pirated torrents. Despite Meta’s reassurance that the company only used publicly available datasets to train its AI models, the team has yet to resolve these issues in court.
Paving the way for future AI innovation
While working on Meta’s Llama AI model seems to benefit those developers’ careers — even if they eventually work for the competition — these retention issues and legal challenges are leading some people to be skeptical about the future of Llama AI.