This week the reassuring headline arrived: AI was oversold, the big firms are hiring back. That's a story about giants — and it's not where this is being felt
A comforting story arrived this week, and I want to be careful with it, because it is half right and the wrong half is doing the work. The story is that AI has been oversold: big companies that cut jobs and blamed the machines are quietly hiring people back, and a large study found the firms that cut hardest are no better off for it. There is a name for the boast now — AI washing. All of that is true. And all of it is about giant companies.
Look at where the evidence comes from. The rehiring examples are the household names. The study surveyed 350 large organisations. The analyst firms that produce this reassurance make their living talking to, and about, the biggest companies in the economy. That is the one part of the labour market loud enough to be measured in real time — and it is not the part where this lands first.
Where it lands first is the bottom, and the bottom is quiet by design. It is the small firm that decides not to replace the person who left. It is the first job — the junior role, the graduate opening — that simply never gets posted. A non-hire holds no press conference. And when we went looking for the number, we found the national statistician had already admitted, in writing, that it does not collect it: the ONS Vacancy Survey asks employers how many vacancies they have, and nothing about whether they are entry-level. The thing the whole country has argued about for a year is not measured.
One figure does get through, from ONS's own count, and it points the opposite way to the reassurance. In the year to this spring, vacancies fell by about 15% in small firms and 13% in the smallest — and rose slightly in the giants. The contraction is at the bottom. The surveys are at the top. That is the whole problem in a line.
So we hold our view, and we sharpen it. Most human work is still heading for the machines. But the comfort being sold this week is read off the one end of the economy big enough to answer a survey, while the end that is actually thinning cannot get itself counted. Do not mistake the silence at the bottom for calm.
This note is mine — the view, the argument, the call to run it. It was drafted with the same AI that writes the rest of the paper, and I put my name to it. And every analyst here runs on models built by Anthropic, a company whose tools sit on both sides of this — sold to the firms doing the cutting, and, on its own account, used for a fraction of what they can already do. We report on this from inside it, and we tell you so.
— M.
—
The Playbook, from the editor:
Distrust the reassurance and the boast equally. "AI took the jobs" and "AI was oversold" are both stories about big companies. Neither describes the small firm or the first job, which is where to actually look.
If you have a school-leaver or a graduate at home, watch the first rung, not the headlines. The jobs going quietest are the ones that never get advertised in the first place.
General information, not financial, legal or career advice.
Reframed after editor push-back: the "AI was oversold / firms are rehiring" reassurance is a large-firm artefact standing in for the whole. The paper's own established evidence (ONS firm-size vacancy split; the ONS FOI blindness) shows the contraction is at the bottom, uncounted. Hold the conviction; correct the record on where it is felt.
Indicts the whole reassurance apparatus — analyst firms and the coverage that repeats them — for reading the labour market off the one end big enough to survey. A competence-and-attention point, not partisan.
- Vacancies and jobs in the UK, June 2026: vacancies down 14.9% (firms 10–49) and 13.3% (1–9) on the year, up 0.9% (2,500+); 707,000 total, lowest since 2021
- ONS FOI: Impact of AI on graduate employment — the Vacancy Survey holds no information on how many vacancies are entry-level
- AI layoffs backfire; Gartner study of 350 large firms; ~32% rehire