Multiverse is a British apprenticeship company founded by Euan Blair, the eldest son of former prime minister Tony Blair. It trains people in AI and digital skills, and most of that training is paid for by the UK government's apprenticeship levy, a payroll tax that funds workforce training.
The company has now started using AI to monitor its own teachers. Every online class is recorded and transcribed, then scored by an AI model for things like slow technical fixes, vague answers, or too many filler words. Managers see a risk score and written notes; teachers do not get the same view.
Multiple teachers describe the effect as constant pressure. They say they now worry more about saying the wrong thing than about actually teaching, and some report losing sleep or seeking therapy since the system began. That is a predictable outcome, not a surprising one.
Research backs this up. Cornell researchers found that AI monitoring tools often lower performance and increase how often people quit their jobs, because being judged by software feels different from being judged by a person. A separate study found employees prefer human oversight over algorithmic oversight, even when the human is stricter.
Multiverse is not the only company testing this. Burger King is piloting AI in staff headsets that counts how often workers say "please" and "thank you" to customers. Meta had to pause an AI program that tracked employee keystrokes and mouse movements after the collected data leaked internally.
Employee monitoring software is growing across industries, and AI is what makes it cheap enough to apply everywhere, not just to call centers. The Multiverse case matters beyond one company because it shows where this technology naturally goes. Once a business can transcribe and score any conversation, it can apply that scoring to any job built around talking to people: sales calls, customer service, training, and coaching.
The apprenticeship levy funding model also means public money is paying for a system that the people delivering the training say is harming them, which raises questions for the public bodies overseeing that spending. Britain's data protection regulator has told employers that workplace monitoring must be necessary, proportionate, and clearly explained to staff in advance, and that automated decisions about people generally need a human check. Multiverse says only human managers write performance reviews and that teachers will eventually see their own data, which keeps it on the right side of that guidance for now.
The lesson for any business thinking about similar tools is simple. AI monitoring that is framed as catching mistakes creates fear and defensiveness. The same technology framed as coaching, with workers seeing their own data first, tends to get accepted instead of resisted.
Companies that skip that distinction are likely to see the same stress and turnover Multiverse is now dealing with, just in a different industry.