In late April 2026, Mark Zuckerberg addressed Meta employees in an all-hands meeting. What he said was so revealing that a recording of the session quickly leaked online.
The message was clear: Meta is systematically using the computer activity of its own highly skilled engineers, their keystrokes, mouse movements, coding patterns, and problem-solving workflows, to train its next-generation AI models. And Zuckerberg was blunt about why they're doing it.
The Core Admission
When asked about employee device monitoring, Zuckerberg explained:
"The AI models learn from having real, from watching really smart people do things. And if you're trying to get it to be able to do certain capabilities, having it be able to observe really smart people doing those things is very important."
He went further, highlighting Meta's unique advantage:
"The average intelligence of the people who are at this company is significantly higher than the average set of people that you can get to do tasks if you're working through these contract companies… So if we're trying to teach the models coding, for example, then having people internally build tools or solve tasks that help teach the model how to code, we think is going to dramatically increase our model's coding ability faster than what others in the industry have the capability to do."
In other words, Meta isn't just scraping public code from GitHub. They're capturing high-quality, real-time behavioral data from some of the best engineers in the world, and feeding it directly into their AI systems.
The Strategic Secrecy
Zuckerberg also addressed why Meta isn't being more transparent publicly:
"We want to communicate as clearly as possible about what we're doing while not having all of the details of things that we think are going to be strategically differentiating leak immediately to the two competitors… Anything that can make the quality of our thing better is generally not something that I think it is in our strategic interest as a company to lay out the details in a lot of detail."
This is the AI arms race in action. Details that give Meta an edge, especially around proprietary training methods, are being kept quiet.
This leaked meeting came just before Meta announced it was laying off approximately 8,000 employees (roughly 10% of its workforce). Many of those affected were told their roles were being eliminated or restructured as the company shifted resources heavily toward AI development.
The pattern is becoming familiar across Big Tech: hire elite engineers, extract their expertise through their daily work, accelerate AI capabilities with that data, then reduce headcount as AI becomes more capable of handling the work.
There's a profound irony here. The very engineers building Meta's empire are, in real time, generating the training data that could make parts of their own job category obsolete. Zuckerberg himself has said in previous interviews that AI will soon handle much of what mid-level engineers currently do.
This isn't science fiction, it's already happening. AI coding assistants are improving rapidly, and companies like Meta are investing tens of billions of dollars annually to accelerate that curve.
This development raises uncomfortable questions:
How much of software engineering will AI actually replace versus simply augment?
Are we entering an era where elite knowledge workers are essentially training their own successors?
What does this mean for compensation, job security, and career planning in tech?
Meta's approach, using internal talent as premium training data, may give them a genuine edge over competitors who rely on lower-quality synthetic data or outsourced labor. But it also highlights a growing tension: companies are incentivised to maximise short-term productivity and shareholder value, even as it accelerates automation of their own workforce.
Mark Zuckerberg isn't hiding the strategy, at least not from his employees. Meta is all-in on AI, and they're using one of their greatest assets (their extremely talented workforce) as fuel for that transformation.
Whether this leads to a golden age of human-AI collaboration or significant displacement for knowledge workers remains to be seen. What's clear right now is that the transition is happening faster than many expected, and some of the best-paid engineers in the world are helping drive the very changes that could shrink demand for their skills.
The age of AI isn't coming. For many at companies like Meta, it's already here.