Co-op Uses AI to Score Legal Advisers Calls Across Britain

Artificial intelligence is intensifying scrutiny of employee performance across multiple industries, with Co-op Legal Services in Britain drawing criticism for using automated technology to evaluate customer calls. The system records conversations between probate advisers and clients, assigns performance scores and gives managers data for assessing staff, prompting concerns about transparency, workplace trust and the limits of automated supervision.

According to reports published by The Guardian and Legal Futures in October 2026, the system uses an OpenAI model to examine more than 50 aspects of each conversation. It generates percentage scores and individual pass-or-fail assessments for managers to use in performance analysis.

The calls involve customers seeking advice about probate, wills and estates, often following a bereavement. Results Sense reported that the monitoring was introduced in the preceding months and involved dozens of employees, with the data also used in efforts to improve sales performance.

A Co-op employee who spoke anonymously to The Guardian described the experience as continuous surveillance rather than occasional compliance checks.

"You are being watched," the employee said.

The worker said the system created distrust because every word and utterance could be assessed for hours each day. The employee also described the technology as "oppressive and dystopian," reflecting concerns that automated assessments could extend beyond conventional workplace quality checks.

Co-op Legal Services has rejected the criticism, arguing that comprehensive call analysis helps maintain service standards and improve staff development. The company said it had received no formal complaints or grievances about the system and that a recent staff survey had not raised negative concerns about it.

A spokesperson for Co-op Legal Services said the technology allowed the company to review every customer conversation rather than relying on traditional sampling methods.

"The efficiency of AI over traditional methods allows us to review all customer conversations, to ensure we consistently deliver high standards of service, while enabling more informed support, coaching and development for colleagues," the spokesperson said.

The company said the assessments could identify effective practices and areas where employees needed additional training. It also maintained that human judgment and accountability remained central to client relationships and that its approach complied with Solicitors Regulation Authority requirements.

The dispute reflects broader concerns about AI-based performance management. A study published by Social Europe, based on a November 2025 survey involving more than 700 workers across UK telecommunications and finance call centers, found that 81.3% of respondents were subject to continuous voice recording, typically incorporating sentiment analysis.

Researchers found that workers questioned whether these systems assessed only the substance of conversations or also evaluated tone, accent and other characteristics. Some reported unexplained declines in performance scores, while union representatives raised concerns about potential disadvantages for neurodiverse employees and workers with particular accents.

According to Social Europe, sentiment-related performance measures could affect variable pay by as much as 40% in some workplaces. Researchers found that employees viewed assessments more favorably when managers could correct errors and workers had meaningful opportunities to challenge scores.

The findings also highlighted concerns about consultation, with employee representatives sometimes involved only after employers had decided to introduce monitoring systems. Researchers argued that stronger human oversight and greater transparency were needed to make automated assessments more accountable.

Regulatory approaches differ across Europe and the United States. Social Europe reported that the European Union's AI Act is scheduled to introduce human-oversight requirements for high-risk applications, including algorithmic management, from December 2, 2027. The United Kingdom, meanwhile, has relied on a principles-based, sector-led approach rather than AI-specific legislation.

In California, Governor Gavin Newsom signed laws in October 2026 restricting certain workplace surveillance practices, according to The Guardian. The measures target practices such as tracking employees' movements and monitoring emotional states, amid wider scrutiny of automated workplace management.

Other companies have also faced questions about employee monitoring. The Guardian reported that technology training company Multiverse had drawn criticism over AI monitoring of online classes, while Meta paused employee keystroke logging during the summer following objections. These cases have added to debate over how employers should disclose monitoring practices and limit the use of automated performance data.