UTAW Releases Report on AI-Driven Work Intensification

On September 24, 2026, the UK tech union UTAW (United Tech & Allied Workers) released a report titled 'AI Workers' Inquiry 2026', which argues that AI redistributes and intensifies work rather than eliminating jobs. Through semi-structured interviews, the report highlights the actual impact of AI adoption on workers in various fields, including software development and data science.

The Reality of AI Adoption: Work Intensification and Deskilling

According to the report, none of the surveyed workers reported a reduction in workload due to AI. Instead, a 'Work Intensification' loop emerged, where expectations for output and review burdens mutually reinforce each other. As LLM generation speed exceeded human review speed, debugging and correction labor increased, but this 'invisible review labor' was not reflected in KPIs.

Concerns about Deskilling and Downskilling were also raised. Junior developers are deprived of foundational learning opportunities and fall into excessive tool dependency, while seniors acknowledge the degradation of their coding skills but resist 'cognitive surrender'. The report also pointed out the issue of expertise gaps being obscured by employers using LLMs as a substitute for employee training.

Token KPIs and the Contradiction of Monitoring

As AI usage became linked to KPIs, compensation, and bonuses, workers faced a contradictory situation where they were disadvantaged for using AI too little or too much. In some organizations, LLM-based evaluators failed to accurately understand human responses, leading to unfair assessments. The report analyzed that such opacity exacerbates information asymmetry in performance reviews and disciplinary procedures.

Policy Proposals: A Binding Regulatory Framework

UTAW demanded a binding regulatory framework through policy proposals covering 32 items across 8 areas. The key contents are as follows:

• Establishment of an AI Regulatory Body: Creation of a statutory agency with the power to investigate and sanction employers and technology providers
• Guarantee of the Right to Refuse: Prohibition of penalties for refusing or criticizing AI use due to ethical, environmental, or safety concerns
• Limitations on Monitoring: Ban on using AI usage data for discipline, promotion, or dismissal, and abolition of AI leaderboards
• Junior Protection: Prohibition of AI replacing education and mentoring, and guarantee of learning time
• Environmental and Health Responsibility: Mandatory impact assessments before AI adoption and legalization of the 'right to disconnect'

The report calls for classifying generative AI as a 'dual-use technology' to strengthen military export controls, and to consider the environmental costs of data centers and human rights issues in the semiconductor supply chain. Unlike corporate data analyses or macro indicators, this document critically analyzes the process of AI adoption within organizations from the workers' perspective, carrying a political character from a union standpoint.