Regulating AI at work in the shadow of a UK election

Why should I read this?

In 2023, the Trades Union Congress (TUC) established a taskforce to draft new legal protections for AI regulation in employment. In April 2024, it published the Artificial Intelligence (Regulation and Employment Rights) Bill. The Bill aims to protect workers where AI systems take or support decisions which pose a high-risk to workers and job applicants.

This is in contrast to the UK government’s current approach. It prefers a lighter-touch, principles-based approach, supported by guidance (e.g. read our Alert on the Recruitment AI Guidance) and enforced by existing regulators such as the Information Commissioner.

While the TUC Bill is not a joint project with the Labour Party, Labour has provided general support and participated in the TUC taskforce. In addition, last year a Labour Party MP sought unsuccessfully to legislate on AI employment rights via a Private Members Bill. Together, this suggests a willingness to legislate in this area, should the Labour Party win the next general election, and the Bill provides some clues (but not a blueprint) as to the potential direction of travel.

What should I do?

Employers should monitor the latest Labour Party election manifesto commitments as they relate to AI at work. Our interactive general election tracker is designed to keep employers updated with the proposed employment policies of the UK's main political parties and will be regularly updated. In terms of mitigating AI workplace risks in the current regulatory environment, our AI at Work Alert sets out key employer actions for consideration, including risk assessments, new policies, refreshing contractual terms and training.

What else do I need to know about the new TUC AI Bill?

According to the TUC taskforce, the Bill is drafted to address “significant gaps” in the existing legal framework. These gaps include: a lack of transparency, observability, explainability and protection against discriminatory algorithms; an imbalance of power over data; and insufficient worker voice and consultation. For example, as a result of the Bill:

• Employers would have to carry out detailed risk assessments of high-risk AI at work decision-making, before and after deployment, followed by yearly rolling reviews. High-risk means having legal or similar significant effects in relation to a worker, employee or job-seeker. The Bill also provides for direct consultation with workers on such risk assessments, to take account of their concerns and interests

• Trade unions (and in their absence, worker representatives) would also have a right to consultation before the deployment of high-risk AI decision making systems and on an annual basis. This duty would be based on the existing collective redundancy information and consultation obligations

• Employers would be required to share certain information in relation to high-risk AI decision-making systems. Examples include: to explain how AI systems operate; to provide a register of systems in operation; to respond to requests from workers for a statement explaining how AI is making high-risk decisions about them; and to provide trade unions with data collected for AI decision making in relation to the union’s members
• Workers would have a right to a human review of high-risk AI decision making made about them
• Employers would be legally incentivised to audit AI systems for discrimination and an employee’s right to claim an automatically unfair dismissal would be extended to include a dismissal where there is unfair reliance on high-risk AI decision-making. New tribunal enforcement provisions and remedies are included in the Bill, including amendments which would make employers expressly liable for discriminatory decisions made by AI systems (with a reversal of the burden of proof making it easier to prove AI based discrimination has taken place)
• Employers would be prohibited from using emotion recognition technology at work

For further details of the Bill, please follow this link.