An algorithm can decide whether you get tomorrow’s shift, and it does not treat everyone the same. Women and migrant workers across the EU face more surveillance, lower pay and less say over automated decisions than their colleagues. Brussels is now drafting rules to catch up.
The scale of the problem is only becoming clear now. An estimated 42.3 per cent of EU workers may already be subject to some form of algorithmic management. That figure could climb to 55.5 per cent within five years. The risks this raises are no longer just theoretical.
For the European Institute for Gender Equality (EIGE), some of the concerns raised several years ago are already a reality. “In 2022, biased algorithms, opaque decision-making and workplace surveillance were often discussed as emerging challenges,” Vytautas Peciukonis, a research officer at EIGE, told EU Perspectives. “Issues such as continuous monitoring, work intensity, stress, reduced autonomy and limited opportunities to challenge automated decisions have therefore moved from largely anticipated risks to documented workplace realities,” he said. The consequences can be costly. Italy’s data protection authority fined Foodinho, a food-delivery platform owned by Glovo, €2.6m in 2021 for failing to ensure its algorithms did not produce discriminatory outcomes.
However, the true size of the problem remains difficult to measure. “Much of the current research does not systematically examine how algorithmic management affects women and men differently, or how gender intersects with migration status, age, disability and other inequalities,” Peciukonis commented.
You might be interested
Algorithms move into offline workplaces
Algorithmic management goes far beyond Uber drivers and food delivery. Online platforms also connect people with jobs in software development, teaching, translation, writing and others. Around 500 digital labour platforms operate in the EU. Precisely how many people depend on them is difficult to measure, but Eurostat estimates around 2 to 5 per cent.
Exposure to algorithm management is increasingly common but uneven across the bloc. At the top are Denmark (70.1 per cent), the Netherlands (65.9 per cent). At the bottom are 26.8 per cent in Portugal and 27.2 per cent in Greece.
“Today, algorithmic management is no longer confined to digital labour platforms,” Peciukonis said. Firms use the tools to allocate schedules, work activities or give workers instructions on how to perform their tasks. European firms used it to monitor working time, whether workers had finished their tasks, and speed. Six per cent used it to monitor the content or tone of conversations, calls or emails.
Woman, migrant, doubly disadvantaged
EIGE warned in 2022 that platform work and AI could create new employment opportunities for women while also reinforcing gender stereotypes, sexism, and discrimination. They found that women earned less than men, particularly those living with a partner and children. Also, the European Parliament argued that women in platform work face greater exposure to algorithmic bias, surveillance, unstable income, and limited opportunities for career progression.
On top of that, foreign-born workers earned around €7,000 less per year. “Migrant women may be particularly exposed because many work in domestic work, cleaning, care and other lower-paid service occupations, including services increasingly mediated through digital platforms,” Peciukonis said. “Migration or residence status, language barriers, non-recognition of qualifications and limited access to trade unions or support services can further reduce workers’ ability to understand their rights or challenge unfair treatment.”
A poor rating, including one potentially influenced by prejudice concerning gender, ethnicity or nationality, may reduce future access to work. — Vytautas Peciukonis, Research Officer, EIGE
If an algorithm uses that rating to decide who receives the next job, human prejudice can have consequences. “When platforms use ratings, rankings or automated systems to allocate tasks and evaluate performance, workers may have little insight into how decisions are made or how to contest them,” Peciukonis said. “A poor rating, including one potentially influenced by prejudice concerning gender, ethnicity or nationality, may reduce future access to work.”
Caretakers offline, workers online
Platforms can be particularly attractive to women trying to combine work with family responsibilities, which remain unequally distributed. Women spent around 2.5 unpaid hours a week on housework and 3.5 more hours on childcare than men. “Algorithms that reward constant availability, rapid responses or uninterrupted working patterns may indirectly disadvantage people with care responsibilities, most of whom are women,” Peciukonis said.
Traditional gender divisions carry over online too. Men dominate technical jobs, while women take on more roles in teaching, writing and translation. “AI and platform work can expand access to employment and flexible earning opportunities, but can also reproduce or deepen existing labour-market inequalities,” Peciukonis told us.
EU rules are trying to catch up
Some protection already exists. The Platform Work Directive, adopted in 2024, gives workers the right to know how an algorithm makes decisions about them, and to have a human review those that matter most. The directive also prohibits platforms from using automated systems to process certain information, including private conversations. This includes data used to infer characteristics such as racial or ethnic origin, migration status, political opinions, disability or health. Member states must transpose it into national law by 2 December 2026. “This can help address worker misclassification and improve access to labour rights and social protection,” Peciukonis commented.
Brussels is now considering whether broader rules are necessary. In July, the European Commission opened the second phase of consultations with employers and trade unions on the upcoming Quality Jobs Act. Algorithmic management and AI at work are among the five areas being considered. The Commission is examining measures to make automated decisions more transparent and human-centred and to protect workers against excessive monitoring. The consultation runs until 28 September.
“The forthcoming Quality Jobs Act provides an important opportunity to extend protection beyond platform work,” Peciukonis said. But the proposal has not yet been published, making it too early to know how far the new protections will go.
For EIGE, shortcomings persist. Existing rules focus mainly on employment status, transparency, data protection and individual rights around automated decisions. Gaps remain in gender-impact assessments, audits for discriminatory outcomes, and sex-disaggregated data. Monitoring of whether algorithms affect women and men differently in recruitment, pay, scheduling, task allocation and career progression is still limited.
