“Rules on the use of artificial intelligence in the labor market, and protection of platform workers” is the third item on the agenda of the Cabinet meeting, held at 11:00 a.m. at Maximos Mansion. This is a bill through which Labor Minister Niki Kerameus is seeking to establish new, clear, and strict rules for the tens of thousands of workers employed through online platforms. The use of artificial intelligence in the workplace has demonstrably created unprecedented challenges that span the entire scope of the employment relationship — and more broadly, the provision of services — from the start and progression of contractual relationships all the way through to their termination.
What many may not have realized is that algorithms — not a company executive — are in many cases shaping the working reality for people who, for example, deliver food, coffee, and other products through popular apps and websites. Algorithms determine the factors and methods of performance evaluation, task assignment, pay structure, and even the criteria for dismissals, thereby amplifying uncertainty and precarity.
As Minister Kerameus explained in an interview with SKAI, “for platform workers, every aspect of their working relationship is defined by an algorithm whose parameters are unknown. With the proposed provisions, the process becomes transparent, while for critical decisions — such as the termination of a working relationship — human oversight will be mandatory.” She further emphasized: “AI is beneficial, but there are also very significant risks. The Greek state is now introducing legislation stemming from a European directive to say: yes to artificial intelligence, it helps us enormously — but with rules.”
Kerameus’s six key interventions on AI use in the workplace
The most significant obligations introduced by the new Kerameus bill, to be presented today at the Cabinet meeting, are as follows:
1. Restrictions on data processing: The bill establishes specific prohibitions on the processing of personal data within the context of algorithmic management. Categories of processing that are explicitly prohibited include:
• Processing of data relating to workers’ emotional or psychological state.
• Processing of data from private conversations and communications.
• Processing of data during periods when the worker is not performing or providing work.
• Processing of biometric data for identity recognition purposes.
• Processing of data to predict trade union activity or the exercise of collective rights.
2. Data protection impact assessment: The bill mandates the obligation to conduct a Data Protection Impact Assessment (DPIA) prior to the introduction or significant modification of automated monitoring or decision-making systems. The DPIA serves as a mechanism for the prior evaluation of risks to workers’ fundamental rights and freedoms. A critical prerequisite for the proper conduct of the DPIA is mandatory consultation with workers’ representatives during its preparation, reinforcing the participatory nature of the assessment.
3. Transparency of automated systems: The bill also establishes general transparency obligations regarding automated monitoring and decision-making systems. Covered businesses are required to provide workers, their representatives, and — upon request — competent national authorities with clear and adequate information about the operation and impact of such systems.
This information includes, among other things: (a) the categories of data and activities monitored, supervised, or evaluated by these systems; (b) the purpose of the monitoring and how it is conducted; and (c) the reasons behind decisions to restrict, suspend, or terminate an account, deny payment, alter contractual status, or any decision with an equivalent or detrimental effect.
4. Human oversight: The bill imposes an obligation to ensure adequate human oversight over automated systems used in the management of service providers. This provision reflects the principle of human-centric artificial intelligence, acknowledging the risks inherent in fully automated decision-making. Specifically, covered businesses are required to designate appropriately trained personnel — with the necessary technical competence and authority — to oversee significant decisions made by or with the support of automated systems. This personnel must be capable of assessing, reviewing, and, where necessary, overriding automated decisions.
5. Review of decisions: The bill guarantees every individual’s right to receive an explanation from a human being regarding decisions that affect them and that have been made or supported by an automated system. This explanation must be comprehensible, sufficient, and relate to the specific circumstances that led to the decision in question. Furthermore, the bill establishes the right to request a review of significant decisions, along with the corresponding obligation to respond to such requests without undue delay and with clear justification.
6. Information and consultation: The bill establishes obligations for informing and consulting workers’ representatives regarding decisions related to algorithmic management. This provision aims to preserve and strengthen the role of collective worker representation bodies in today’s modern digital work environment. Consultation is not a mere formality — it must take place in a timely and substantive manner, with the genuine aim of enabling workers’ representatives to influence the final decision.
Finally, subcontracting chains present a particular challenge, creating multi-layered contractual relationships that make it difficult to identify who bears responsibility for compliance with workers’ rights. For this reason, the bill provides for the joint and several liability of all parties in the subcontracting chain for the effective protection of individuals performing work through intermediaries, given that these individuals also fall within the scope of the algorithm’s operation.
