In September 2023, the TUC set up a taskforce with the AI Law Consultancy at Cloisters Chambers, and the Cambridge University Minderoo Centre for Technology and Democracy, in order to manage the drafting of the Artificial Intelligence (Regulation and Employment) Bill (“The Bill”).
The Bill was drafted by Robin Allen KC and Dee Masters of the AI Law Consultancy at Cloisters chambers. The TUC was assisted in administration of the project by the Cambridge University Minderoo Centre for Technology and Democracy.
The Bill benefitted from the input and expertise of a Special Advisory Committee, which met three times across 2023 and 2024.
Members of the Committee included representatives from a diverse range of organisations and political parties, including the Ada Lovelace Institute, the Alan Turing Institute, Connected By Data, TechUK, UKBlackTech, the British Computer Society, CIPD, the RAI UK, Cambridge University, Oxford University, Prospect, Community, CWU/UTAW, USDAW, GMB and cross-party MPs. The policy expressed in the Bill is that of the TUC and should not be taken to express the policy of these organisations unless explicitly stated.
The Bill regulates the use of artificial intelligence systems by employers in relation to workers, employees and jobseekers to protect their rights and interests in the workplace.
The Bill also provides for trade union rights in relation to the use of artificial intelligence systems by employers, addresses the risks associated with the value chain in the deployment of artificial intelligence systems in the field of employment, and enables the development of safe, secure and fair artificial intelligence systems in the employment field.
The rights and obligations contained in the Bill will be enforceable in the Employment Tribunal which is ordinarily a ‘no cost’ jurisdiction (where the parties are responsible for their own costs regardless of outcome).
Authored by Robin Allen KC and Dee Masters, AI Law Consultancy at Cloisters Chambers
The Bill is divided into thirteen Parts, each covering a different aspect of artificial intelligence and employment, with four Schedules which complement these Parts. There are also Explanatory Notes.
Part 1: Preliminary
This part introduces the structure of the Bill.
Part 2: Core Concepts
This part of the Bill defines the core concepts that define the remit of the Bill such as ‘artificial intelligence system’, ‘high-risk decision-making’, ‘data’, ‘processing’, ‘emotion recognition technology’, ‘employee’, ‘worker’, ‘jobseeker’ and ‘employer’.
The decision by an employer or its agent to deploy artificial intelligence systems for ‘high-risk decision-making’ is the trigger for most of the rights and obligations in the Bill.
‘Decision-making’ means any decision made by an employer or its agent in relation to its employees, workers or jobseekers taken or supported by an artificial intelligence system.
Decision-making is ‘high-risk’ in relation to a worker, employee, or jobseeker, if it has the capacity or potential to produce legal effects concerning them, or other similarly significant effects.
The term ‘jobseeker’ is new and covers ‘a person who is actively seeking new employment, whether or not that person is already employed’.
The definition of ‘employer’ includes an existing employer of a worker or employee and a prospective employer in relation to a jobseeker.
Part 3: Transparency, Observability, and Explainability
This Part enacts positive duties on employers and their agents in the employment sphere.
A new type of assessment called a ‘Workplace AI Risk Assessments’ (WAIRA) is created.
An employer cannot undertake high-risk decision-making until a WAIRA has risk assessed an artificial intelligence system in relation to health and safety, equality, data protection and human rights.
There will need to be direct consultation with employees and workers before high-risk decision-making occurs; the WAIRA will be central to that consultation.
Employees will need to establish and maintain a register of information about the artificial intelligence systems used in high-risk decision-making.
There will be a right to personalised explanations for a high-risk decisions which are or might reasonably be expected to be detrimental to employees, workers or jobseekers.
Employees, workers or jobseekers will be entitled to a right to human reconsideration of a high-risk decision.
Part 4: Prohibition on Detrimental Use of Emotion Recognition Technology
This part prohibits the use of emotion recognition technology in high-risk decision-making that may be detrimental to a worker, employee or jobseeker.
Part 5: Prohibition on Discrimination
Existing rights in the Equality Act 2010 are amended to tailor them to the use of artificial intelligence systems by employers and their agents in relation to employees, workers and jobseekers.
The amendments include that employees will not be liable for the discriminatory consequences of artificial intelligence systems used by their employers, and employers will need to prove that systems are not discriminatory in order to avoid liability subject to a new audit defence.
This new audit defence will allow employers and their agents to successfully defend a discrimination claim where they did not create or modify the artificial intelligence system and conducted thorough auditing before deploying it, including introducing careful procedural safeguards.
Part 6: Health and Wellbeing
There is a statutory right to disconnect which will be added to the Employment Rights Act 1996.
Part 7: Dismissal
This part states that it will be automatically unfair dismissal to dismiss an employee through unfair reliance on high-risk decision-making or as a punishment because an employee has exercised their right to disconnect. The right to interim relief pending determination of any complaint will also be extended to these scenarios.
Part 8: Trade Unions
There are provisions for the fair use of data so that trade unions can be provided with the data collected by employers in relation to their members.
Existing collective consultation obligations in relation to trade unions are also extended to situations in which an employer is proposing to do high-risk decision-making.
The consultation must begin at least one month before the high-risk decision-making takes place and must be repeated every 12 months for as long as decision-making continues.
The consultation must include consultation about the risks to the rights of employees and the measures envisaged to address the risks.
Part 9: Auditing and Procedural Safeguards
This section of the Bill sets out the auditing of artificial intelligence systems for discrimination, and the standards which an employer must meet to rely on the auditing defence set out in Part 5.
Part 10: Regulators and Bodies in the Employment Field and Artificial Intelligence
This part sets out regulatory obligations concerning artificial intelligence and details the principles that key regulators (identified within Schedule 3) must apply in any context concerning employment and the deployment of artificial intelligence systems.
This paper sets out the background to the TUC Artificial Intelligence (Employment and Regulation) Bill (“the Bill”), the multi-stakeholder process behind the drafting, why the Bill is needed, and how it could improve the rights of working people.
AI is rapidly transforming our society and the world of work, yet there are no AI related laws in place in the UK, nor any current plans to legislate soon.
Urgent action is needed to ensure that people are protected from the risks and harms of AI-powered decision making in the workplace, and that everyone benefits from the opportunities associated with AI at work. Employers and businesses also need the certainty offered by regulation.
And the more say workers have in how technology is used at work, the more rewarding and productive the world of work will become for us all.
British workers are overwhelmingly supportive of more worker consultation, with 69% of working adults in the UK agreeing that employers should have to consult their staff first before introducing new technologies such as AI in the workplace. 1
The Bill translates many of the principles and values that seem to attract near universal support (such as the importance of consultation, transparency, explainability and equality) into concrete rights and obligations. It represents a significant step forward in the movement towards the responsible adoption of AI.
2010 to address the use of artificial intelligence systems by amending the burden of proof and introducing a new defence for employers and their agents, where they have audited the system for discrimination.
(1) The Core Concepts in this Act are those defined in this Part.
(2) Cognate phrases to those defined in this Part are to be construed accordingly.
(3) Regulations made under, and guidance and codes published in accordance with, the powers in this Act are to be construed accordingly.
In this Act -
In this Act “emotion recognition technology” means an artificial intelligence system used in whole or in part for the purpose of identifying or inferring the attention, emotions, or intentions of natural persons on the basis of their biometric data.
In this Act -
In this Act a “trade union” has the same meaning as section 1 of the Trade Union and Labour Relations (Consolidation) Act 1992.
and such regulations may make different provision for different circumstances.
and such regulations may make different provision for different circumstances.
and such regulations may make different provision for different circumstances.
No high-risk decision-making using emotion recognition technology may be used which is, or might reasonably be expected to be, to the detriment of a worker, employee, or jobseeker.
“CHAPTER 1A
Artificial Intelligence at Work
“(aa) in identifying B as a jobseeker;
(ab) advertising to B as a jobseeker”
“(1A) Any decision-making done by an artificial intelligence system which is deployed by an employer must be treated as done by the employer.”
“(2A) Any decision-making within the meaning of the Artificial Intelligence (Regulations and Employment Rights) Act 2024 deployed by an agent for a principal, with the authority of the principal, shall be treated as also done by the agent and the principal.”
Section 110 Equality Act 2010 is amended to add after section 110 (3) –
“(3A) A does not contravene this section if A is an employee and he relied to any extent on decision-making within the meaning of the Artificial Intelligence (Regulations and Employment Rights) Act 2024 which is deployed by an employer.”
Section 124 Equality Act 2010 is amended to add after (3) –
“(3A) Where the contravention relates to the use of discriminatory artificial intelligence systems, in setting out an appropriate recommendation, the tribunal must have regard to Part 9 of the Artificial Intelligence (Regulation and Employment Rights) Act 2024.”
“(3A) Where person (A) is alleged to have contravened Part 5 (work), as a result of reliance on an artificial intelligence system within the meaning of the Artificial Intelligence (Regulations and Employment Rights) Act 2024, unless A shows that the provision was not contravened (whether by A or by the artificial intelligence system), the court must hold that the contravention occurred.
(3B) If A cannot discharge the burden of proof set out in subsection 3A, it is nevertheless a defence to a claim under this Act where A is the employer or its agent, and
(3C) If A successfully relies on the defence in section 3B, this does not preclude other persons in the artificial intelligence value chain from being liable under the Equality Act 2010 beyond Part 5 (work)”
Schedule 25 to the Equality Act 2010 (Information Society Services) is amended by the deletion of paragraphs 1 and 2.
The Employment Rights Act 1996 is amended by the insertion after section 63K of a new section as follows:
“PART 6B
STATUTORY RIGHT TO DISCONNECT
63L. Statutory right to disconnect
63M. Enforcement
and, in the absence of evidence establishing the contrary, an employer, shall be taken to decide on a failure to act when he does an act inconsistent with doing the failed act or, if he has done no such inconsistent act, when the period expires within which he might reasonably have been expected to do the failed act if it was to be done.
63N. Remedy
“47H. Right to disconnect
“104H Unfair high-risk decision-making
“(f) the rights conferred by ss 14 to 18 and 21 the Artificial Intelligence (Regulation and Employment Rights) Act 2024”.
“(gn) section 104H applies”.
“104I Right to disconnect
“(go) section 104I applies”.
“(3A) Where the dismissal is unfair pursuant to section 104H, the tribunal may make a recommendation to the employer to ensure that there is no repetition in accordance with Part 11 of the Artificial Intelligence (Regulation and Employment Rights) Act 2024”.
“(5) Where the tribunal makes an award of compensation because the dismissal is unfair pursuant to section 104H, it will not bar the employee from seeking a remedy for the infringement of their rights and entitlements under other legislation, but the principle of no double recovery for the same loss shall apply.”
Section 128(1)(a)(i) of the Employment Rights Act 1996 is amended by substituting “, 103A or 104H” for “or 103A”.
“Section 134B
“Chapter IIA Artificial Intelligence Systems
198C Duty to consult representatives
and shall be undertaken by the employer with a view to reaching agreement with the appropriate representatives,
the employer shall be treated as complying with the requirements of this section in relation to those employees if he complies with those requirements as soon as is reasonably practicable after the election of the representatives.
198D Election of representatives
198D Complaint and compensation
198D Definitions
and such regulations may make different provision for different circumstances.
Microbusinesses
The Secretary of State may by order disapply or modify the obligations in sections 14, 15, 16, 17, 18 and 38 for any employer who employs fewer than 10 employees.
Subordinate legislation
Final provisions
There is to be paid out of money provided by Parliament any increase attributable to this Act in the expenses of a Minister of the Crown.
This Act forms part of the law of Great Britain [note it can easily be adapted to extend to Northern Ireland, however there are further and different provisions in Northern Ireland relating to the protection of human rights and prohibition of discrimination].
This Act may be cited as the Artificial Intelligence (Regulations and Employment Rights) Act 2024.
Want to hear about our latest news and blogs?
Sign up now to get it straight to your inbox
To access the admin area, you will need to setup two-factor authentication (TFA).