Ke Yang, the executive Apple had just tapped to lead its nascent ChatGPT-style search effort inside the company, is leaving to join Meta — a sudden exit that exposes how fiercely Big Tech is fighting for AI talent and raises new questions about Apple’s roadmap for a revamped Siri.

What happened

Yang had been quietly elevated in recent weeks to run a group at Apple called Answers, Knowledge and Information (AKI), the team charged with building the web-connected, conversational layer Apple hopes will make Siri act more like a modern generative assistant that can pull up-to-date facts from the internet. People familiar with the matter told Bloomberg that Yang will depart for Meta after only a short stint as AKI’s leader.

Why it matters

AKI isn’t a side project — it sits at the center of Apple’s publicly stated plan to overhaul Siri next spring. The new Siri is intended to combine Apple’s emphasis on user privacy with much richer, generative capabilities: longer-form answers, web sourcing, and deeper use of on-device personal context to handle complex requests. Losing the executive who was visible internally as the point person for that work is, minimally, a distraction; at worst, it could slow a high-profile product objective timed for March.

The talent bleeds

Yang’s departure is the latest in a steady trickle — and sometimes a flood — of departures from Apple’s AI ranks this year. Apple’s foundation models group has already seen a number of senior researchers and engineers leave, including Ruoming Pang, the group’s founder and lead scientist, who moved to Meta earlier this summer. Several others have followed Pang’s route, joining Meta’s newly assembled Superintelligence Labs or other startups and rivals.

A cluster of recent moves fills out the picture: Sam Wiseman, a New York–based Apple researcher, tweeted that he joined Reflection AI earlier this month; Chong Wang — described in reporting as one of Apple’s more senior researchers — reportedly left for Meta recently; and Frank Chu, who led infrastructure and search-related AI efforts at Apple, moved to Meta over the summer. Taken together, those moves illustrate both Meta’s appetite and its ability to convert Apple’s in-house experience into its own projects.

Meta’s strategy: build fast, buy people

Meta has been unusually aggressive in staffing up a high-ambition AI group and has repeatedly surfaced as the destination for senior hires from Apple’s AI teams. Those hires are concentrated in a few areas: foundation models, training and infra, and search/answers work — all skills that are immediately useful to a firm trying to scale generative services and internal model training. That combination of people and infrastructure is precisely what makes poaching attractive to Meta and makes retention an urgent problem for Apple.

Who’s in charge now (for the time being)

Internal reporting lines appear to be shifting: several outlets report that Apple is moving the AKI group under Benoit Dupin, a senior director who oversees machine-learning-related cloud infrastructure and reports to John Giannandrea, Apple’s head of AI and machine learning. Dupin’s background — including prior roles touching search and cloud systems — makes him a logical caretaker for the effort; whether he will be a permanent replacement or a stopgap is unclear. Apple has not issued a public comment about Yang’s departure.

What this does to Apple’s AI story

From a product narrative standpoint, Apple has been trying to walk a tightrope: deliver generative features on par with OpenAI and Google while maintaining the privacy posture that differentiates Apple. Executives and engineers who can stitch model training, on-device inference, and web grounding together are scarce. When they leave en masse, the obvious risk is slower progress or feature-scope tightening as teams refocus. That said, Apple still has institutional advantages — hardware, a massive installed base, and partnerships with cloud vendors — that mean a stalled timeline isn’t the same as failure. Bloomberg’s reporting makes clear the company is also doing behind-the-scenes recruiting and organizational work to steady the ship.

Signals vs. noise: how worried should anyone be?

It’s tempting to read every exit as a sign of doom, but there are two counterpoints. First: talent flows in Silicon Valley are dynamic — offers, counteroffers, and mission mismatches happen all the time. Second: the complexity of modern AI programs means leadership changes don’t always translate immediately into product collapse — teams can reassign responsibilities, hire replacements, or change scope. Still, the clustering of senior departures to one competitor is unusual and costly, and it raises a real question about morale and momentum inside Apple’s AIML organizations.

Ke Yang’s move to Meta is a sharp, symbolic reminder that the AI fight is now as much about people as it is about models or chips. Apple’s ambitions for a smarter, web-aware Siri were already a tall order; the company will need to move decisively to hold the ground it’s claimed. For consumers, the short-term outcome is likely to be one of two things: a slightly delayed feature rollout or a narrower initial feature set that Apple iterates on more slowly but with tighter privacy controls. Either way, the month ahead will say a lot about whether Apple can turn headcount churn into a manageable reorganization — or whether the company will fall further behind firms that are spending aggressively to assemble model, infra and search muscle in one place.