Artificial intelligence is changing how British employers assess performance, monitor productivity and evaluate customer interactions. While businesses argue that these systems can improve quality assurance and efficiency, growing concerns over employee privacy, transparency and accountability are raising questions about how far workplace monitoring should go.
The debate intensified after Co-op Legal Services introduced an artificial intelligence system to analyse employees’ telephone conversations with customers seeking advice on wills, probate and estates. According to The Guardian, the system assesses more than 50 aspects of each call and generates scores that managers can use to evaluate performance.
The Co-op says the technology helps improve service quality and supports staff development, rather than making decisions independently. However, some employees have raised concerns about the intensity of the monitoring, particularly when conversations involve customers experiencing bereavement.
The case highlights a broader question for British workplaces: how can employers use AI to improve performance without undermining workers’ privacy, trust and ability to challenge assessments?
How AI is changing employee monitoring
Traditional workplace monitoring often involved periodic performance reviews, supervisor observations and occasional checks on customer interactions. AI-powered systems can analyse much larger volumes of information, including telephone calls, computer activity, attendance records and digital communications.
These tools can identify patterns, flag potential problems and generate performance scores. Employers may use the findings to improve training, identify compliance issues or maintain consistent service standards.
However, automated assessments can also oversimplify complex human interactions. An employee handling a distressed customer, for example, may need to prioritise empathy and careful explanations rather than follow a standard conversational pattern.
The Co-op case illustrates this tension. Assessing more than 50 aspects of a call may provide managers with detailed feedback, but the number of criteria alone does not establish whether the system accurately measures empathy, professional judgement or the quality of support provided to customers.
According to The Guardian’s reporting on 8 October, the technology is used to review customer conversations and produce assessments for managers. The Co-op has defended its approach, saying that human judgement and accountability remain central to its service.
What does British data protection law require?
The Information Commissioner’s Office (ICO), the UK’s data protection regulator, states that employee monitoring must be lawful and fair. Employers must identify an appropriate legal basis for processing personal information, explain their monitoring practices clearly and ensure that the measures are proportionate to their intended purpose.
The regulator also warns that excessive monitoring can intrude on workers’ private lives, undermine trust and adversely affect mental wellbeing. These risks can be particularly significant for remote workers, whose personal environments may be inadvertently captured by monitoring systems.
The ICO’s guidance does not prohibit workplace monitoring. Instead, it requires employers to balance legitimate business interests against workers’ rights and consider whether less intrusive methods could achieve the same objective.
For example, an employer seeking to improve customer service might initially use targeted quality checks and structured coaching rather than automatically analysing every interaction. The appropriate approach will depend on the purpose, risks and circumstances of the monitoring.
The legal position becomes more complex when an AI-generated assessment influences a consequential employment decision.
Under Article 22 of the UK General Data Protection Regulation, certain solely automated decisions that have legal or similarly significant effects on individuals are restricted. The ICO explains that meaningful human involvement requires more than a manager simply approving a computer-generated score. Decision-makers should be able to examine the evidence, interpret the recommendation and reject it when appropriate.
This distinction is important: an AI-generated performance score is not automatically unlawful, but employers must consider how personal information is processed and how the resulting assessment is used.
The evidence: What do the case study and research tell us?
The Co-op example provides a concrete illustration of how AI monitoring is being applied in a British workplace. Yet one company’s experience cannot establish how widespread such systems are across the UK economy.
A further indication of the policy debate comes from the Chartered Institute of Personnel and Development (CIPD), the professional body for human resources and workplace development.
In its response published on 30 September 2026 to the government’s consultation on workplace monitoring technologies, the CIPD supported the introduction of a statutory code of practice. It argued that such a code could clarify expectations under existing law and help organisations adopt monitoring technologies responsibly.
The CIPD also warned that non-statutory guidance alone would not provide an enforceable framework for responsible adoption.
The government consultation, which closed on 30 September, sought views on how workplace monitoring could be made fairer and more transparent while ensuring that workers’ rights keep pace with technological change.
These developments demonstrate that workplace AI monitoring is not simply a question of whether a particular tool works. It is also a question of what safeguards employers should provide and how existing legal protections should operate in increasingly automated workplaces.
An important evidence gap remains: the sources cited here do not establish a reliable, representative percentage of British employees currently subject to AI-based performance monitoring. Further national research would help policymakers understand the scale of adoption and whether particular groups of workers face greater risks.
Experts call for accountability and meaningful oversight
The concerns extend beyond individual employers. Trade unions have questioned whether automated assessments risk placing too much authority in systems that employees may struggle to challenge.
In The Guardian’s report, John Chadfield, a national officer at the Communication Workers Union, criticised the growing reliance on automated workplace assessment, arguing that “Unaccountable computer systems should not be people’s managers.”
His comments reflect a wider concern: workers should not have to accept an assessment simply because it has been generated by a sophisticated algorithm.
The CIPD’s policy response offers a practical route towards addressing these concerns through clearer standards. A statutory code could help organisations understand what responsible monitoring should involve, although its eventual impact would depend on the rules adopted and how effectively they are implemented.
For employees, meaningful safeguards would include clear explanations of what is monitored, why information is collected, how performance scores are calculated and who can access the results. Workers should also have a way to correct inaccurate information and challenge assessments they believe are unfair.
For employers, transparency can help maintain confidence in performance management while reducing the risk of disputes arising from unexplained or disproportionate monitoring.
Can AI improve productivity without damaging employee wellbeing?
The benefits of AI monitoring should not be dismissed. Used appropriately, it can help identify training needs, highlight recurring service problems and reduce the administrative burden associated with reviewing large volumes of customer interactions.
However, monitoring every interaction can create pressure if employees feel that each pause, phrase or difficult conversation may count against them. This concern is particularly relevant in emotionally demanding roles, where the quality of an interaction cannot always be captured by numerical indicators.
A responsible approach should therefore combine automated analysis with contextual judgement. Managers should examine whether a score reflects genuine performance problems or circumstances that the system cannot adequately understand.
Employers should also assess whether monitoring is achieving its stated purpose. If a system is introduced to improve customer service, organisations should evaluate customer outcomes and employee feedback alongside productivity figures.
The question is not simply whether AI can measure more activity. It is whether the information it produces leads to fairer assessments and better decisions.
What happens next?
The UK’s debate over workplace monitoring is entering an important policy phase. The government’s consultation and the CIPD’s call for a statutory code indicate growing attention to the relationship between technological innovation and employee protection.
Any future framework will need to address transparency, proportionality, data protection and the role of human decision-makers. It will also need to distinguish between monitoring that supports workers and systems that exert excessive control over their daily activities.
The Co-op Legal Services case shows why these questions matter in practice. AI can provide detailed information about workplace performance, but detailed measurement does not necessarily mean a fair or complete assessment.
For British employers, the challenge is to demonstrate that monitoring systems serve a legitimate purpose and that their benefits outweigh the risks. For workers, the central issue is whether they can understand, question and challenge the systems being used to evaluate them.
As AI becomes more deeply embedded in employment practices, maintaining that balance will be essential to building workplaces that are both technologically capable and respectful of individual rights.
