Union Demands AI Negotiation to Protect Mental Health at Work

On November 19, the General Council of Psychology of Spain hosted the III Mental Health at Work Conference, where labour unions pushed for the formal negotiation of artificial intelligence systems in workplaces.

The confederal health and safety department of the General Workers’ Union argued during the event that algorithmic tools function directly as work organization decisions and psychosocial risk factors. Víctor Sanz Sáez, a technician for the union’s health department, spoke on a panel addressing priorities for a human-centric digital transition.

He shared the discussion with labour law professor Adrián Todolí Signes, Vicente Sánchez Jiménez representing business organization CEOE, and Jorge Alonso del Rosario representing self-employed association ATA, according to Uso.

USO’s representatives maintained that companies must evaluate and prevent algorithmic impacts at their inception under the existing Law on Occupational Risk Prevention. Traditional risk assessments often fail to address automated systems that can instantly alter work rhythms, tasks, or shift schedules. Sanz Sáez pointed out that human supervision remains scarce and optional in current practices, creating what the union calls an invisible risk.

Legal mandates for human oversight will not take effect until December 2, 2027, when the European Artificial Intelligence Regulation comes into force. Until then, current rules offer workers only a right to information rather than collective bargaining rights over automated parameters.

Did You Know? Only 0.8% of pathologies reported through the PANOTRATSS system are classified as psychosocial in origin, even though mental health issues represent the second leading cause of temporary disability in Spain.

Mental health damage rarely appears in official workplace injury registries because current occupational disease tables do not include psychosocial disorders. Union officials outlined several specific mechanisms through which algorithmic management harms worker well-being. These include constant digital surveillance, work intensification, reduced autonomy, system opacity, unpredictable shift changes, inherent data biases, and automated disciplinary sanctions.

Article 64.4.d of the Workers’ Statute currently requires companies to inform worker representatives about algorithmic parameters, but it excludes those parameters from mandatory collective bargaining. To address these gaps, the union outlined several key demands for upcoming workplace negotiations.

The union’s proposal calls for specific risk assessments before any algorithmic system goes live, alongside mandatory bargaining over parameters that influence worker health. Additional requirements include guaranteed human supervision, strict limits on work intensity and schedule predictability, and better official recognition of psychosocial injuries.

They also requested specialized training, clear protocols for labour inspectors, and fully funded expert advisory services for union sections.

The III Mental Health at Work Conference was held on November 19, hosted by the psychology council’s work and organizational psychology division.

Article 64.4.d of the Workers’ Statute obligates companies to inform employee representatives about the parameters used by workplace algorithms.

The General Workers’ Union (UGT), founded on 12 August 1888 in Mataró by Pablo Iglesias Posse, emerged from secrecy during Spain’s democratic transition after Franco’s death in 1975. Alongside Workers’ Commissions (CCOO), it constitutes one of the two major avenues for workers’ representation in modern Spain, growing from 200,000 members in 1920 to over one million by 1932 before the Civil War forced it into exile and clandestinity for nearly four decades.

The European Artificial Intelligence Regulation is scheduled to enter into force on December 2, 2027.