AI Godfather, Yann LeCun, a towering figure in AI and Meta’s departing chief AI scientist,  has left the company with a scathing assessment of its strategy and leadership, sounding alarms about the future direction of one of the world’s most influential tech giants.

In an explosive interview with the Financial Times, LeCun pulled no punches, highlighting manipulated benchmarks, internal strife, and a mass exodus of top AI talent following the release of Meta’s latest language model, Llama 4.

LeCun, who spent over a decade shaping Meta’s AI research, paints a picture of an organisation at war with itself- torn between scientific integrity and commercial ambition. The crux of the controversy erupted after the April 2025 launch of Llama 4, where LeCun alleges that performance benchmarks were “fudged” to exaggerate the model’s capabilities. He claims that different models were cherry-picked for specific tests, a revelation that eroded trust at the highest levels and sent shockwaves through the AI division.

The fallout was swift and severe. Meta CEO Mark Zuckerberg, reportedly furious over the misleading data, lost confidence in the entire generative AI team, effectively sidelining them and triggering a significant departure of talent. “A lot of people have left, a lot of people who haven’t yet left will leave,” LeCun revealed.

Compounding the turmoil was Meta’s $14 billion acquisition of Scale AI, elevating its young founder Alexandr Wang to lead the new TBD Lab and placing him above LeCun in the hierarchy. LeCun, never one to mince words, bristled at the move: “You don’t tell a researcher what to do,” he said.

LeCun has long argued that large language models – like those championed by Meta – are a dead end on the path to true superintelligence. He described a company culture increasingly dominated by a “LLM-pilled” mindset, unwilling to explore alternative approaches. For LeCun, scientific integrity trumped corporate pressure: “My integrity as a scientist cannot allow me to do this,” he insisted.

Frustrations had been simmering for years, as LeCun’s research teams saw their innovative ideas sidelined in favour of safer, more conventional projects. “When you do that, you fall behind,” he warned, likening Meta’s internal environment to a political battleground stifling true progress.

Having departed Meta late last year, LeCun is now charting a new course as executive chair of Advanced Machine Intelligence Labs. His new venture is laser-focused on developing “world models” powered by V-JEPA, a novel architecture designed to learn from video and spatial data rather than just text. LeCun’s vision – Advanced Machine Intelligence (AMI) – aims to build systems capable of genuine reasoning, planning, and understanding of the physical world, staking out a bold alternative to today’s AI orthodoxy.

With early technology demonstrations expected within a year and the promise of full-scale systems on the horizon, LeCun remains undaunted by the challenges ahead. “Maybe there is an obstacle we’re not seeing yet, but at least there is hope,” he concluded – leaving both Meta and the wider AI community to ponder what comes next.

His influence continues to grow beyond his own company. Earlier this week, US-based AI startup Logical Intelligence announced LeCun as the founding chair of its Technical Research Board. Even world leaders are paying attention: French President Emmanuel Macron reportedly texted LeCun after news of his departure broke.

RELATED STORIES

AI titan Yann LeCun and ex-deputy PM Nick Clegg, team up to drive Europe’s AI ambitions

AI pioneer Yann LeCun quits Meta following reported rift with Zuckerberg