Sam Altman has claimed that Meta is offering OpenAI staff bonuses worth up to $100 million in a bid to boost its artificial intelligence ambitions, but insists none of his top talent has taken the offer.
The OpenAI CEO made the remarks during a podcast that Meta, the parent company of Facebook, Instagram, and WhatsApp, is aggressively scaling up its AI efforts following a substantial investment in the data labelling firm Scale AI.
Meta has reportedly committed nearly $15 billion to the San Francisco-based startup as part of a push to train next-generation AI models.
Altman acknowledged Meta’s interest in recruiting from OpenAI, but brushed it off. “So far none of our best people have decided to take them up on that,” he said.
“I’ve heard that Meta thinks of us as their biggest competitor. Their current AI efforts have not worked as well as they have hoped and I respect being aggressive and continuing to try new things.”
Meta confirmed its partnership with Scale AI in a statement to Reuters, saying the company will play a central role in training its future AI systems.
“We will deepen the work we do together producing data for AI models and Alexander Wang will join Meta to work on our superintelligence efforts,” the company said.
Wang is the founder and CEO of Scale AI and will now contribute to Meta’s most ambitious AI projects.
Altman, however, was skeptical of Meta’s approach. “I think that there’s a lot of people, and Meta will be a new one, that are saying ‘we’re just going to try to copy OpenAI’,” he said.
“That basically never works. You’re always going to where your competitor was, and you don’t build up a culture of learning what it’s like to innovate.”
The comments come amid the intensifying rivalry among leading tech firms in the race to build and deploy large-scale AI systems. While OpenAI, backed by Microsoft, has gained significant ground with products like ChatGPT and partnerships across industries, Meta has been relatively quieter in comparison - until now.
The investment in Scale AI marks one of the largest commitments to AI infrastructure outside of model training. Data labelling is a crucial component in developing high-quality foundation models, particularly in domains such as computer vision and language understanding.
Despite the bold moves, industry observers say Meta faces an uphill battle in competing with OpenAI, Anthropic, and Google DeepMind, all of whom have years of experience in deploying advanced models. Meta has previously open-sourced its LLaMA models in an effort to gain developer traction, but has yet to launch a major consumer-facing success on the scale of ChatGPT.
RELATED STORIES
Meta lifts CapEx to $72bn as AI ambitions fuel data centre push
