Credit: © u_c3dlob4ax4 / Pixabay via Canva.com An OECD employer survey of 6,047 firms across six countries found that 67% of US firms use software to sanction poor performance, against 4% in the four European countries surveyed. American firms also monitor the content and tone of conversations at 55% against 6%, and the OECD attributes the gap to regulatory architecture. Most of these tools are not AI, which the OECD warns leaves them outside rules written specifically for it. The OECD surveyed 6,047 firms across six countries on whether they use software to instruct, monitor or evaluate workers.
The fieldwork was done by Ipsos between June and August 2024, with mid-level managers answering for workplaces of 20 or more people. Adoption is high almost everywhere: 90% in the United States, 81% in France, 78% in Germany and Spain, 76% in Italy and 40% in Japan. What separates the countries is not whether firms use these tools but what they use them for. The evaluation gap is enormous Software that sanctions poor performance is used by 67% of US firms, 4% in the four European countries surveyed and 1% in Japan.
Software that rewards good performance runs at 83% against 13%. Performance leaderboards visible to other staff are in half of US workplaces and 7% of European ones. Any evaluation tool at all: 90% in the US, 35% in Europe, 11% in Japan. Monitoring splits the same way.
US firms track the speed of work at 72% against 15% in Europe, and monitor the content and tone of conversations, calls or emails at 55% against 6%. Intensity, not just presence The OECD measured 15 distinct use cases. More than three quarters of American firms use ten or more of them, and very few use fewer than eight. European firms typically use three to five.
Nearly a third of Japanese adopters use exactly one. So the headline adoption numbers flatten a real difference. A French firm with a rota tool and an American firm running conversation analysis, speed tracking and automated sanctions both count as adopters. That distinction is missing from most of the debate. Arguments about what AI is doing to European jobs tend to focus on displacement rather than on how the people still employed are managed.
The OECD is explicit about why It attributes the difference to regulatory architecture, contrasting the centralised, rights-based approach of the European Union with a patchwork of agency enforcement and state and local rules in the United States. Consultation law does specific work here. An EU directive requires information and consultation of worker representatives, while US bargaining obligations attach only to particular subjects. The effect shows in who gets asked.
Among US firms that consult at all, almost all managers say they were consulted and about a third say other employees were. Italy and Spain outperform France and Germany This is the finding European policymakers should sit with. All four operate under the same GDPR, yet Italy and Spain have markedly more governance measures in place, with audits at 83% and 78%. The OECD credits specific national law: Italy’s Transparency Decree of 2022, which explicitly covers automated decision-making, and Spain’s 2021 Riders’ Law, which gives works councils a right to be informed about algorithmic management.
Enforcement mattered too. The Foodinho case, in which Glovo’s Italian subsidiary was fined 2.6 million euros under data protection law, sits behind those audit numbers. The AI Act problem Most of this is not artificial intelligence. The OECD says plainly that algorithmic management tools vary in sophistication and are not necessarily AI-powered, and warns that rules written specifically for AI leave gaps open for tools that do not use it.
The software managers actually named is the ordinary enterprise stack: SAP, Workday, Oracle, Jira, Asana, Trello, and time-tracking products. Because it runs on systems firms already own, the OECD notes it is often not recognised or branded as algorithmic management at all. That is a scoping problem rather than an enforcement one. TNW has examined what the AI Act does and does not do for jobs, and a rule keyed to AI cannot reach a punch-clock that quietly acquired a scoring function. Workers may not know Ninety-one percent of managers say employees or their representatives are made aware of the software.
The OECD immediately doubts it, suggesting most managers are unaware of how little workers know. A related finding is harder to explain. Sixty-nine percent of European managers say their tools do not collect or use data on individual workers, which the OECD calls anomalous and attributes to managers not understanding the tools or to answering carefully under GDPR. Where managers do acknowledge data use, most say workers cannot opt out.
In the US that reaches 90%, with more than half saying workers cannot request corrections. What the survey cannot tell you No workers were surveyed. Every figure on satisfaction or stress in the OECD’s own data describes managers reporting on their own jobs, and the report says so. Response rates ran from 2% in Germany to 6% in Japan, which the OECD states plainly.
The definition of adoption is also broad enough to include data collection with no algorithmic processing at all. European here means France, Germany, Italy and Spain, not the EU. Rules on workplace monitoring vary considerably across the bloc. What to watch Watch the Commission. The OECD notes the debate reopened after von der Leyen called for an initiative on algorithmic management, and researchers have argued for extending the Platform Work Directive beyond platforms.
Watch whether transparency duties get teeth. The AI Act already obliges employers to support AI literacy among staff operating these systems, and more employment obligations land this year. Neither helps if the tool is not classed as AI.
Leave a Reply