In April 2026, Reuters reported that Meta was installing software on U.S. employees’ computers to capture mouse movements, clicks, and keystrokes for training its AI models. The story reached HR Voices through 1Password, where Weston Fillman and Gabrielle Caron lead two halves of the People function. Wes runs people operations and employee relations, and Gab leads talent, culture, and growth. They sent host Rebecca Taylor a Business Insider article about it, and she used it to open a conversation about employee monitoring and trust.
Neither guest defended the rollout, and neither dismissed the idea behind it. Wes pointed to mandatory participation with no opt-out. Gab’s first reaction was empathy for leaders under pressure to know who is productive. The choice, in Gab’s words, “probably had good intent, but debatable on the execution.”
Why the Tool Was Not the Problem
Wes did not call the tool a mistake. Training a model on how people work is not necessarily, in his words, “the wrong thing for an AI company to do.” The failure he saw was a communication breakdown around the why. Without a reason he could explain transparently, he said, he would face “a huge loss of trust” with his own teams.
Gab went further. Explained properly and handled with care for privacy, the idea could be inspiring, which is what made the execution a “missed opportunity.” The cost is steep at a privacy and security company. “That is the foundation of our product,” Gab said of trust.
Rebecca raised a second cost. A traveler told to walk normally past a security dog forgets what normal is, and employees who know they are monitored may keep to what they think is expected. Research published in Harvard Business Review found that monitored employees were substantially more likely to break rules, including purposely working at a slow pace.
Gab expects the data to suffer too. People may be gaming the clicks and jobs change fast, so a model trained this way is likely to “hit a plateau.”
How Change Management Became the Rollout Itself
A leader recently asked Gab where the change management strategy for AI was. The reply was “we are the change management strategy.” The older model was “a paper or a framework” that arrived stage-gated after the announcement. The replacement fits in one line. “It has to be the project with the humans.”
The familiar parts survive. Gab still wants FAQs, office hours, and training sessions. What changes is the timing: leaders communicate the change while they make it, because there is “no time to do it on the side.”
Many employers skip the explaining. Gallup reported in 2025 that 44 percent of U.S. employees said their organization had begun integrating AI, while only 22 percent said it had communicated a clear plan.
1Password shows what the explaining buys. Gab said the company started on some enterprise AI tools six to nine months before we recorded and now sits at 98 percent adoption, the first of the HR metrics the People team tracked. Its latest experience survey averaged 70 percent and above on whether AI helps people do a better job, with some dimensions lower than others.
What Actually Works
Start as the User
Gab’s first tip for an HR leader is to “familiarize yourself with the tool.” A less technical People leader’s own experience sets a fair bar for how the average employee receives it. Where a change can be piloted, tracking included, Gab wants leaders to go through it themselves.
Gab’s summary was blunt: “be the user. You have to know this stuff.” Quick wins inside the People team are the subject of our crash course on how HR teams can start adopting AI.
Tie the Tool to the Mission and the Workday
The second move is the why, told from more than one angle, since motivations differ. Gab ties AI to what 1Password does. The company secured passwords, then devices, and now has to secure agents, which it cannot do without using them.
Wes works at the level of the workday. His own adoption deepened when he fed meeting summaries into an internal tool and asked what he had committed to that day. He has seen one person go from scared to intermediate user “in just a few short weeks.”
Give People Room to Experiment Out Loud
The third move is structure, meaning access, training, open channels, and time. Wes hosts a weekly session for the People partners and operations team, where anyone can bring a problem or an idea. The rule is that “no question is a dumb question,” which gives a busy team the psychological safety to experiment.
Rebecca called it learning out loud, failures included. Recognition follows the same logic. 1Password is building AI into performance review conversations by asking how the tools helped employees reach their goals instead of checking how they used them.
Where Employee Relations Fits
Wes runs the employee relations function as well as people operations. He starts with the concerns, not the pitch. He has acknowledged his reports’ worries about the environmental and social impact of AI, and told us he shares some of them. Then he turns to how the work could improve.
At 1Password, the rule for the gaps is candor. When leaders do not have an answer, they say so and state the intent.
ER teams bringing AI into their own casework face the same test. Our AI agent for employee relations keeps the team reviewing every step. Rebecca gave the reason in passing, calling trust “one of the hardest things to earn and one of the easiest things to break.”
Frequently Asked Questions About AI Rollouts and Employee Trust
The missing explanation, in the guests’ reading. Wes described tools that tracked keystrokes, mouse movements, and screen activity to train an AI model, with mandatory participation, no opt-out, and a breakdown in communicating the why.
Not by itself, in Gab’s view. Jobs change fast and people may game the clicks, so the approach is likely to plateau. Knowing which tools people use, and when, has more value for security.
Three moves ran together. Leaders used the tools first, tied them to the company mission and to each person’s work, then added training, FAQs, open forums, and space to experiment.
It runs inside the project instead of after it, in Gab’s account. Leaders communicate a change while they make it, with FAQs and office hours still in place, because there is no time to bring people along later.
Usage first, then what the saved time becomes. Gab asks whether people feel better about their output and can reinvest the time in learning or different work. Without that, the gains plateau.
The Bottom Line for HR Leaders
Weston Fillman and Gabrielle Caron’s framing moves the Meta story away from the tool and onto the rollout. In their reading, a tool with a defensible purpose reached a workforce without the reason attached, and trust lost that way is hard to win back. Their alternative is not slower. It is the same project with the people inside it from the start.
1Password treats its 98 percent adoption as a starting line. Wes pitches the work to his teams as future-proofing, a form of upskilling for current employees that serves them wherever they work next.
Wes placed the moment in a longer arc, from the personnel department through the push into people analytics to AI. A function that understands people and data, Gab concluded, is better prepared for what follows.
See how AllVoices helps People teams bring AI into employee relations work without losing employee trust.
