Workplace monitoring is no longer limited to CCTV, access cards or checking whether employees are logged into their computers. Employers increasingly use software to track digital activity, assess productivity, allocate shifts, identify risk and support HR decisions.

Artificial intelligence is accelerating that change. An AI system may summarise performance data, identify patterns in absence, flag possible misconduct or recommend disciplinary action. The technology may improve efficiency, but it can also create serious legal and employee-relations risks if workers do not understand how it operates or cannot challenge its conclusions.

The government’s Make Work Pay: workplace monitoring technologies consultation is now open. It was published on 8 July 2026 and closes at 11:59pm on 30 September 2026. The consultation is not yet law, and the government has not selected a preferred approach. However, its proposals indicate that employers may face greater expectations around transparency, worker consultation, fairness and human oversight.

This article explains what employers should know now.

What counts as workplace monitoring technology?

The consultation uses a deliberately broad definition. Workplace monitoring technologies (WMT) are digital tools used to collect, track, analyse or make decisions based on information about workers and their activities.

Examples include:

• GPS and location tracking;
• keystroke logging and time-on-application monitoring;
• email and communications monitoring;
• CCTV and individual video monitoring;
• biometric access systems, including facial recognition and fingerprint scanning;
• automated productivity or performance scoring;
• algorithmic scheduling and work allocation;
• health, fatigue or physiological monitoring; and
• AI systems that inform or make decisions affecting workers.

The definition is important because AI is only one part of the proposed framework. An employer does not need to use generative AI for the consultation to be relevant. A conventional monitoring system can still create substantial legal risk if it influences performance management, pay, promotion, discipline, redundancy or dismissal.

The consultation also distinguishes between algorithmic management and solely automated decision-making. Algorithmic management may provide recommendations or scores to a manager. Solely automated decision-making involves an outcome being reached without meaningful human involvement.

What is the government consulting on?

The consultation presents three broad options. These could potentially be combined, and none should be treated as a final policy decision.

1. A statutory code of practice

The first option is a statutory code of practice supported by more detailed, non-statutory guidance.

The code would set expectations for the responsible introduction and use of WMT. It would not create a standalone tribunal claim or replace existing employment, equality or data protection duties. However, an Employment Tribunal could take the code into account when considering an existing claim, such as unfair dismissal or discrimination.

The consultation proposes that compensation could potentially be adjusted by up to 25% where an employer unreasonably failed to follow the code and the claimant succeeded in a relevant claim.

For example, suppose an employer introduces keystroke monitoring without clearly explaining its purpose, assessing whether it is proportionate or giving workers a meaningful opportunity to raise concerns. If monitoring data is later relied upon to dismiss an employee, the tribunal may consider the employer’s approach when assessing an unfair dismissal or discrimination claim.

A statutory code would therefore create a clearer benchmark for good practice, even if it did not itself create new rights.

2. A statutory duty to consult and negotiate

The second option would be more significant for employers. It would introduce a legal requirement to consult and negotiate, with a view to agreement, with recognised trade unions or elected staff representatives before introducing WMT or making significant changes to existing systems.

Agreement would not necessarily have to be reached. The proposed duty is about giving workers and representatives a genuine opportunity to understand the proposal, identify risks and influence the outcome.

The consultation asks detailed questions about:

• which technologies should trigger consultation;
• whether the duty should apply to all new systems or only higher-risk systems;
• whether significant changes in purpose or use should trigger a fresh process;
• whether the duty should cover performance, discipline, pay or dismissal decisions;
• what information employers should provide;
• how long consultation should last; and
• what remedies should apply if the process is not followed.

Possible remedies include protective awards, financial compensation, a requirement to provide information or a requirement to delay implementation until a compliant process has taken place.

This approach could be particularly challenging where software is updated frequently, introduced in stages or supplied by a third party. Employers would need to understand not only the system’s original purpose but also how later changes affect workers.

3. Non-statutory guidance

The third option is non-statutory guidance alone. This would be the most flexible and least burdensome approach.

Guidance could include sector-specific examples, checklists, case studies and practical tools covering procurement, data protection, worker engagement, human oversight, accuracy and ongoing review. It could also be updated more easily as technology changes.

The limitation is that guidance would not itself create new legal obligations. Its effectiveness would depend heavily on voluntary compliance and whether employers already take responsible monitoring seriously.

The eight principles employers should prepare for

Although the final outcome is unknown, the consultation identifies eight principles that provide a useful framework for reviewing current practices:

1. Purpose and rationale – employers should be able to explain why monitoring is necessary.
2. Transparency and understanding – workers should understand what data is collected and how it may affect them.
3. Worker engagement and voice – employees and representatives should have a meaningful opportunity to raise concerns.
4. Fairness and equality – monitoring must not produce discriminatory or disproportionate outcomes.
5. Necessity, proportionality and privacy – employers should consider whether a less intrusive method would achieve the same objective.
6. Human oversight and accountability – managers must remain responsible for significant decisions.
7. Dignity and wellbeing – employers should consider the effect of monitoring on stress, autonomy and workplace culture.
8. Accuracy, reliability and review – systems should be tested and reviewed regularly.

These principles are consistent with existing obligations. The consultation is not a substitute for compliance with the Equality Act 2010, UK GDPR, the Data Protection Act 2018, employment contracts or the Acas Code of Practice.

The ICO’s guidance on monitoring workers explains that monitoring should be lawful, fair and transparent. Intrusive or high-risk monitoring may require a Data Protection Impact Assessment. Workers should also receive clear information about the purpose of processing, the data collected and how it will be used.

AI-generated grievances are creating a new HR challenge

The impact of AI is not limited to employer monitoring. Employees are also using generative AI to prepare grievances, disciplinary responses and workplace complaints.

An AI-generated grievance remains a grievance. An employer should not dismiss or ignore it simply because the language appears unusually polished, lengthy or legalistic. The underlying concerns must be identified and handled under the employer’s normal procedure.

The Acas guidance on disciplinary and grievance procedures confirms that employers should follow fair procedures, investigate relevant issues and provide an opportunity for the employee to respond and appeal where appropriate.

In practice, employers should:

• acknowledge the complaint promptly;
• identify the employee’s actual allegations and desired outcome;
• arrange a meeting to clarify unclear or exaggerated wording;
• check factual assertions rather than accepting AI-generated legal references;
• investigate relevant evidence fairly;
• avoid assuming that an AI-assisted complaint is dishonest; and
• keep clear records of the investigation, meetings and outcome.

Employees should not be penalised merely for using AI to help express a concern. However, employers can still address genuinely abusive, knowingly false or vexatious conduct where the evidence supports doing so.

Employers should also be cautious about using AI to draft grievance outcomes. Confidential personal information, witness evidence and sensitive employment data should not be entered into an external AI system without proper safeguards. Any assistance with summarising information should remain subject to human checking and decision-making.

Why AI decisions must remain fair and non-discriminatory

An AI system can reproduce or amplify bias in the data used to train or operate it. A productivity tool may disadvantage part-time workers, disabled workers or employees with caring responsibilities if it treats constant availability as a measure of commitment. A recruitment system may produce less favourable outcomes for particular groups if its historical data reflects past discrimination.

Employers remain responsible for the consequences. It is not a defence to say that “the algorithm made the decision”.

Before relying on an AI tool in an employment decision, employers should ask:

• What data does the system use?
• Is the data accurate and relevant?
• Could the system disadvantage people with protected characteristics?
• Has the system been tested using representative data?
• Can a manager explain the recommendation?
• Can the employee correct inaccurate information?
• Is there a genuine route to human review or appeal?
• Is the proposed use consistent with the original purpose for collecting the data?

Human oversight must be meaningful, not a rubber stamp. A manager should have authority, time and information to question an AI recommendation. If the manager simply approves whatever the system produces, the process may still be effectively automated.

These issues can become particularly important in discrimination, whistleblowing, grievance and dismissal cases. Employees who believe monitoring has produced unfair treatment may raise internal complaints or bring tribunal proceedings. Our existing post on Employment Discrimination: What UK Employees Need to Know About the 2026 Equality Act Claims Surge explains the importance of evidence, internal grievances and strict tribunal time limits. Tyndel Solicitors also advises on whistleblowing and protected disclosures, including situations where a worker suffers detriment after raising concerns about wrongdoing or unlawful workplace practices.

Practical steps for employers in August 2026

Employers do not need to wait for the consultation outcome before taking sensible action.

1. Create a monitoring and AI register

Record every system that collects or analyses worker information, including tools operated by suppliers. Note its purpose, data sources, users and potential effect on employment decisions.

2. Review policies and privacy notices

Policies should explain what is monitored, why monitoring takes place, how long information is retained and how workers can challenge inaccuracies. A separate AI policy should address acceptable employee use, confidentiality and HR decision-making.

3. Carry out equality and data protection assessments

Consider a DPIA where monitoring is intrusive or high-risk. Assess whether the system could create direct or indirect discrimination and whether reasonable adjustments are needed.

4. Introduce human review safeguards

No significant decision should rest solely on an unexplained AI score. Require a named decision-maker to review the context, check the evidence and record the reasons for the final outcome.

5. Consult early

Even before a statutory duty exists, consulting employees or representatives is prudent. Early engagement can expose inaccurate assumptions, reduce resistance and create a clearer audit trail.

6. Train HR and managers

Managers should understand that AI outputs are evidence to assess, not conclusions to adopt automatically. They should also know how to handle AI-generated grievances under the Acas framework.

7. Preserve evidence

Keep records of procurement decisions, testing, consultations, impact assessments, system changes and human review. These records may become important if a worker challenges a disciplinary, redundancy, pay or dismissal decision.

What should employers do before 30 September?

The consultation is open to employers, workers, trade unions, professional bodies and other interested parties. Responses can be submitted through the official GOV.UK consultation page.

Whether the government chooses guidance, a statutory code, a legal consultation duty or a combination of options, the direction of travel is clear: employers will be expected to justify workplace monitoring, communicate openly and retain responsibility for fair outcomes.

If you need advice from employment law solicitors UK on workplace monitoring, discrimination, grievances or AI-related HR decisions, early advice can help identify risks before they become disputes. If your employment relationship is ending or you have been offered terms to resolve a dispute, a settlement agreement solicitor UK can review the agreement, advise on your rights and help negotiate appropriate terms.

Contact Tyndel Solicitors’ Employment Law team for practical advice and representation in England and Wales.

This article provides general information only and does not constitute legal advice. The consultation proposals may change, and employers should obtain advice on their specific circumstances.