A Meta executive who has been overseeing the firm’s internal AI restructuring programme is leaving the company, according to Reuters.
The news agency cited an internal announcement posted on 17 June which said Emily Dalton Smith, previously head of product for Meta’s microblogging site Threads, will be leaving the firm.
Dalton Smith is understood to have been spearheading efforts to consolidate Meta’s internal AI tools into Metamate, its internal enterprise AI assistant which employees can use to analyse data, schedule meetings, and summarise documents.
“Our goal is to make Metamate the starting point for all kinds of work – from doing deep research to prototyping a new feature to putting together a sales presentation,” Dalton Smith wrote, in a May memo to Meta employees seen by Reuters.
She added that her team intended to pull in functionality from AI systems that can generate code based on conversations, navigate work files and retain “persistent memory” of people’s work, she said.
In her internal announcement, Reuters added Dalton Smith said she was committed to staying at Meta and work with the firm’s chief technology officer Andrew Bosworth until a transition to “what’s next” had been completed.
No precise timeline for Dalton Smith’s departure has been set publicly.
Dalton Smith has been at Meta since 2015, having led its social good and social impact product teams for six years before shifting to product management in 2021.
Meta has spent the past six months increasingly focusing its workforce on AI efforts. At its @Scale conference in January, Meta reported that 50 per cent of its workforce use Metamate daily and in May, Reuters reported that Meta was reassigning 7,000 employees to AI-focused teams.
That move came at the same time as the tech giant made 8,000 other workers redundant.
After acquiring a 49 per cent stake in Scale AI in June 2025, Meta onboarded its former chief executive Alexander Wang to lead its AI team and restructured its AI teams to AI projects by use case.
This saw it create four departments covering AI assistants, infrastructure, long-term breakthroughs, and undisclosed research.
At the time, publications such as Bloomberg reported that executives at Meta felt its Llama 4 models had not met expectations and was working to deliver competitive LLMs. The firm ultimately scrapped its would-be frontier model Llama 4 Behemoth.
Critics of the new approach include Yann LeCun, who had served as Meta’s chief AI scientist and led its AI research team since 2013. He left the firm in January citing disagreements with the firm’s strategy. In January, LeCun told the FT that LLMs a “dead end when it comes to superintelligence”.
Since the restructuring, Meta has released a new LLM family for its social media platforms, Muse Spark.
