Co-op Deploys OpenAI to Score Bereavement Calls and Squeeze Probate Staff

Imagine losing your mother, picking up the phone to sort out her estate, and pouring your grief out to a legal adviser. Now imagine an OpenAI model sitting silently on the line, transcribing every pause, weighing vocal cadence, and grading the clerk on fifty distinct metrics before spitting out a percentage score.

That is not a thought experiment. That is Tuesday morning at Co-op Legal Services.

What happened

The British retail and services giant has placed dozens of its legal services employees under continuous, automated audio surveillance. Workers fielding probate calls are having their entire shifts recorded, transcribed, and evaluated by software running an OpenAI large language model.

Every exchange regarding deceased estates, wills, and estate administration is run through an automated rubric. The system breaks each call down into more than fifty separate components, issuing pass or fail ratings alongside aggregate percentage marks that land directly on managerial dashboards.

Legal advisers handling probate files spend up to seven hours a day tied to headsets. Under this setup, virtually every second of that customer-facing time is subject to algorithmic review.

Internal whistleblowers report that the software does not merely inspect legal accuracy. It also grades the employee on sales efficiency, identifying missed opportunities to cross-sell additional commercial products to callers navigating legal bureaucracy after a death.

The Co-operative Group has defended the deployment, asserting that the model serves purely as an operational aid rather than an automated executioner. Executive leadership claims the AI is not making firing decisions, but rather giving team leaders the data needed to train advisers to communicate with uniform empathy.

The office is turning into a digital assembly line where an algorithm scores your tone while you talk to an orphan.

The background

Co-op is an unusual corporate creature. Founded on nineteenth-century mutualist ideals, the Manchester-based group runs grocery chains, funeral homes, insurance underwriting, and legal practices. Its consumer branding rests almost entirely on ethics, community ownership, and social purpose.

That ethical veneer is running straight into standard platform capitalism. Previously, quality assurance in high-volume contact centres relied on spot checks. A human manager would sample three or four calls a month, flag statutory compliance lapses, give feedback, and move on.

Automating that oversight changes the math completely. Machine listening removes the cost barrier to total surveillance. Instead of sampling five percent of a shift, an employer can monitor one hundred percent for pennies per hour of audio compute.

Co-op is hardly the pioneer here, but it is moving aggressively. Consider the corporate pattern established throughout 2026:

• In February 2026, Burger King rolled out acoustic tracking across American locations to monitor customer interactions at the counter.
• Euan Blair’s apprentice training firm, Multiverse, faced heavy pushback after deploying persistent AI scrutiny across remote classroom lectures.
• Meta attempted to roll out continuous employee keystroke logging during the summer of 2026, only to back off after an intense staff revolt and regulatory scrutiny.
• The International Labour Organization issued warnings regarding the psychosocial dangers of algorithmic workplace observation, citing elevated stress and depression.

The Co-op rollout marks the arrival of continuous acoustic evaluation inside white-collar, emotionally sensitive legal practice.

What people are saying

Workers subjected to the tool describe the working environment as oppressive. Staff members speaking anonymously note that knowing an algorithm tracks every breath erodes morale. The constant surveillance creates a permanent layer of distrust between workers and management.

Union leadership is mobilizing against the software. John Chadfield, national officer at the Communication Workers Union, condemned the deployment in stark terms:

Unaccountable computer systems should not be people's managers, and these hard-working employees demand far more respect.

Chadfield argues that treating AI as a universal operational cure degrades workplace dignity and artificially intensifies routine tasks without offering any tangible benefit to the public.

The Chartered Institute of Personnel and Development views the transition as structural. Hayfa Mohdzaini, senior policy adviser at the institute, observed that performance reviews are adopting the mechanics of the factory floor, introducing structured metrication to roles that previously relied on discretion.

Co-op leadership insists the panic is misplaced. Caoilionn Hurley, managing director of Co-op Life Services, maintained that the deployment ensures high service standards across vulnerable interactions. Hurley argued that the tooling guarantees clients receive consistent, compassionate guidance, maintaining that human accountability remains the foundation of their legal client relationships.

Corporate spokespeople flatly reject the whistleblower claims of low morale, countering that probate advisers remain engaged and that the system acts strictly as an educational baseline for internal coaching.

What happens next

The immediate consequence will be labour friction. Frontline legal workers did not train in estate law to have their bedside manner graded by a predictive text engine. When you take people whose job requires emotional resilience and turn their conversations into scorecards, they quit. The Co-op will likely face accelerated attrition among experienced probate handlers who refuse to work under a machine microscope.

Watch the legal definition of managerial surveillance closely. The International Labour Organization has highlighted the absence of enforceable international frameworks governing generative models used as supervisory proxies. The UK Information Commissioner’s Office and worker tribunals will eventually have to answer whether continuous emotional analysis of calls constitutes disproportionate workplace intrusion under existing privacy regimes.

There is also a commercial trap. Co-op’s brand equity depends on perceived decency. When consumers realize their moments of raw grief are being parsed by an OpenAI instance to maximize sales performance, the reputational fallout will dwarf whatever minor operational efficiencies the model delivers.

Algorithmic empathy is an oxymoron. You cannot automate genuine human sympathy, but you can certainly use it to turn an office into a sweatshop.