By Veronica Nilsson, General Secretary of the Trade Union Advisory Committee to the OECD (TUAC), and PierPaolo Bombardieri, General Secretary of the Unione Italiana del Lavoro (UIL).
From mechanisation to electrification to digitalisation, workers have had to organise and fight for their share of every technological revolution. AI is the latest, and it is already reshaping workplaces on an extraordinary scale: in three out of four Italian firms, algorithms are used to manage aspects of work. Across Europe the figure is 79%; in the United States, 90%.
Algorithmic management systems assign tasks, monitor and evaluate performance, and filter job applications, often without workers knowing what data is being used or having any means to challenge the outcome. But that is only one dimension of the challenge. Translators’ incomes are falling as their craft is reduced to reviewing and refining text generated by a machine, while screenwriters have had to fight to prevent studios from using AI-generated scripts as a way to undercut their pay. Workers across different sectors are seeing their work become more intense and less engaging, with their roles requiring an increasing amount of AI-checking and supervisory tasks at a pace dictated by machines. Others discover that the work they produced is being used to train the systems that may replace them. These are all decisions about the organisation of work – and the organisation of work is a matter for collective bargaining.
Where collective bargaining is absent, workers are shut out of these decisions entirely. The OECD’s own survey found that while 63% of firms claim to consult workers on AI tools, those consultations primarily target managers – the people operating the systems, not the people subject to them. In practice, this means workers often find out about AI-driven decisions after they have been made, if they find out at all, while the productivity gains flow to employers and the adjustment costs fall on them. When machines produce more, workers should work less and live better. But that only happens when unions are at the table to negotiate it.
Where collective bargaining is present, the outcomes are different. In the United States, the Writers Guild of America negotiated landmark protections on AI-generated content and intellectual property, while ZeniMax workers represented by CWA secured advance notification requirements for AI deployment and limits on how AI can be used. In Ireland, the Financial Services Union reached an agreement with the Bank of Ireland on AI in 2025. CCOO and UGT in Spain negotiated an agreement with Just Eat that created an Algorithm Commission with union representation. In Nordic countries, the OECD’s own data shows that where workers have real influence over how AI systems are introduced, the harmful effects are significantly reduced. The evidence is consistent: collective bargaining produces better AI governance.
At the international level, the OECD is one of the key institutions shaping AI governance. Its policy guidance influences how governments across 38 member countries approach the regulation of AI and its impact on workers. TUAC brings the voice of trade unions into OECD policy-making, working to ensure the interests of workers are reflected in high-level guidance and recommendations. These include the OECD AI Principles, which call on governments to ensure AI respects labour rights and to use social dialogue as part of ensuring a fair transition for workers, and the Due Diligence Guidance for Responsible AI, which provides a framework for companies to identify and mitigate risks throughout AI’s supply chain.
The G7 Action Plan on AI in the World of Work, agreed under the Italian presidency in 2024, further calls on governments to promote social dialogue and collective bargaining at all levels in the development, deployment and monitoring of AI. TUAC is pushing the OECD to build on these commitments and more strongly recognise the issues AI raises for workers and the role of collective bargaining in addressing them. And everywhere, unions are on the front line to govern – not simply endure – the impact of AI on workers.
UIL’s Padua Charter, adopted at its XIX National Congress in July 2026 and presented to the Italian government and employers, sets a comprehensive framework for governing AI’s impact on work. It demands that employers negotiate with unions before introducing any algorithmic system, maintain a register of all algorithms in use, and provide workers with a comprehensible explanation of any automated decision that affects them. Beyond transparency, it sets hard limits: AI and automation cannot be used as grounds for dismissal, and when AI increases productivity, the gains must be shared through reduced working hours at equal pay rather than captured as profit while workers absorb the transition costs. It also insists that training on AI – including critical thinking, source verification and the recognition of bias – must be an enforceable right delivered during working hours, not a discretionary benefit.
To ensure that these commitments translate into practice, UIL has established ONIL, an observatory on AI, work and rights, to track what actually happens when AI systems enter workplaces – the kind of monitoring that should inform AI governance at every level, including at the OECD. As UIL’s Padua Charter recognises, regulation writes the rules, but collective bargaining is what makes them real in the workplace.
These examples demonstrate that trade unions are prepared and forward-looking, not only assessing the impacts of AI on workers today, but anticipating future scenarios and their implications. This experience and foresight should be fully recognised by governments and relevant international organisations, including the OECD, by ensuring that trade unions are meaningfully involved in shaping major decisions on the future development and governance of AI.
This article is also available in Italian on UIL’s Terzo Millennio: https://terzomillennio.uil.it/blog/come-la-contrattazione-collettiva-sta-definendo-le-regole-dellia-sul-posto-di-lavoro/
