Blog 09 - Ethics, Privacy and Employee Trust in AI: How Much Workplace Data Is Too Much?

Ethics, Privacy and Employee Trust in AI: How Much Workplace Data Is Too Much?

Modern organisations can collect more information about employees than at any previous point in the history of work. Digital systems can record working time, task completion, location, communication patterns, customer ratings, productivity, response speed and even indicators connected with health, fatigue or alertness. Artificial Intelligence (AI) can then analyse these data to support managerial decisions.

From a management perspective, this creates obvious opportunities. Better information can support workforce planning, identify risks, improve safety and make decisions faster. From an employee perspective, however, the same technology can create a very different feeling: “My employer can see almost everything I do.”

The ethical question is not simply whether an organisation can collect employee data. It is whether collecting and using those data is necessary, fair, transparent and respectful of employee dignity.

The Rise of the Algorithmic Manager

The International Labour Organization defines algorithmic management as the use of algorithmic systems and tracked data to organise, assign, monitor, supervise and evaluate work. Importantly, not every algorithmic-management tool is a sophisticated AI system. What they share is the ability to automate or support activities that were traditionally performed by human managers.

The OECD’s cross-country employer survey shows how normal this has already become. Among surveyed organisations, 90% of managers in the United States said their firms used at least one algorithmic-management tool to instruct, monitor or evaluate workers. The average was 79% across France, Germany, Italy and Spain, while adoption in Japan was 40% (OECD, 2025).

Figure 1. Adoption of algorithmic-management tools across surveyed countries. Author’s illustration based on OECD (2025).

The OECD identifies examples including tools that monitor work completion, working time, work speed, the content or tone of calls and emails, location, fatigue, health and safety. Evaluation tools can also set targets, reward performance, sanction poor performance and maintain performance leaderboards.

This makes workplace privacy an HRM issue rather than merely an IT-security issue. These technologies affect the employment relationship, managerial power, employee autonomy and the degree of trust between workers and organisations.

When Does Measurement Become Surveillance?

Organisations have legitimate reasons to measure work. Employers need to know whether employees attend work, whether customers are being served properly and whether health and safety standards are being followed. The ethical difficulty arises when monitoring becomes excessive or extends beyond what is reasonably required for the job.

For example, measuring whether a delivery was completed may be directly relevant to performance. Tracking an employee’s precise location throughout every minute of the day is a much more intrusive practice. Reviewing customer-service quality may be legitimate. Analysing the emotional tone of every conversation raises different questions about privacy and proportionality.

The OECD’s trustworthy-AI framework emphasises that workplace AI should respect fundamental rights such as privacy, fairness and labour rights. It also argues that employment-related AI decisions should be transparent and understandable by humans, with clear accountability when something goes wrong (OECD, 2023).

More data is not automatically better HRM.

An organisation may improve one performance metric by intensifying monitoring while simultaneously weakening autonomy, trust and the employment relationship.

Trust Changes Depending on What AI Is Allowed to Do

Employee trust is not simply a choice between “trusting AI” and “not trusting AI”. People appear to make distinctions based on how much authority technology is given.

A CIPD poll conducted in January 2025 received 2,214 responses. It found that 63% would trust AI to inform important work decisions, while only 1% would trust AI to make important work decisions. The CIPD therefore emphasised a human-centred approach, clear organisational guidelines, data security, ethical practice and human oversight.

Figure 2. Trust in AI as decision support compared with AI as decision-maker. Author’s illustration based on CIPD (2025).

This suggests that many employees may accept AI when it acts as a decision-support system but become uncomfortable when responsibility appears to disappear into the technology.

Imagine an AI system warning a manager that an employee may be at risk of leaving. The manager can use that signal as a reason to have a conversation. Now imagine the system automatically deciding that the same employee should be denied promotion because the model predicts a high probability of resignation. The data may be similar, but the ethical significance is completely different.

Trust is easier to maintain when AI provides evidence and people remain visibly accountable for decisions that affect other people.

The Power Problem: Who Controls Employee Data?

Workplace data are not collected within an equal relationship. Employers usually have greater power to decide which technologies are introduced, what is measured and how information is used.

Clegg, Courpasson and Phillips (2006) emphasise the importance of power in organisational life. Digital monitoring provides a modern example. When employers can observe employees continuously while employees cannot see how algorithms evaluate them, an information imbalance develops.

Transparency therefore becomes more than a technical feature. Employees should know:

• What information is being collected?
• Why is the information required?
• How long will it be retained?
• Who can access it?
• Is AI analysing the information?
• Does it influence pay, performance, promotion or discipline?
• Can the employee correct inaccurate information?
• Can a human review an automated recommendation?

Without answers to these questions, employees may comply with a monitoring system while still distrusting the organisation behind it.

A Real Regulatory Boundary: Emotion Recognition at Work

The European Union’s AI Act provides a powerful example of how ethical concerns are becoming legal boundaries.

The European Commission lists emotion recognition in workplaces and education institutions among the AI practices prohibited by the Act, subject to the regulation’s specific exceptions and definitions. These prohibitions became applicable in February 2025 (European Commission, 2026).

This is important because emotion-recognition systems attempt to infer emotional states from signals such as facial expressions, voice or behaviour. In an employment setting, using such systems could affect how employees are judged while relying on highly sensitive interpretations of human behaviour.

The AI Act therefore illustrates a broader ethical principle: not everything that technology can infer about a worker should automatically become legitimate management information.

Accountability Cannot Be Outsourced to Software

The OECD’s 2025 employer survey found that nearly two-thirds of managers using algorithmic-management tools had at least one concern about their impact on workers. The most frequently reported issue was unclear accountability when an algorithmic decision was wrong, followed by difficulty understanding the logic of decisions and concerns about worker health.

This raises a simple but important question:

If an AI-supported employment decision harms an employee, who is responsible?

“It was the algorithm” is not an acceptable management philosophy. HR professionals and managers remain responsible for deciding whether a system is appropriate, whether its output makes sense and whether the final decision is fair.

Employee Data Should Have a Clear Life Cycle

A human-centred approach to employee data should begin before information is collected. Organisations should first identify a legitimate purpose and then ask whether the same goal can be achieved with less intrusive data.

Figure 3. Human-centred workplace AI data cycle. Author’s illustration informed by OECD, CIPD and EU AI governance principles.

A responsible HR approach should therefore include:

• Purpose: define why employee data are required before collecting them.
• Data minimisation: collect only the information necessary for that purpose.
• Transparency: clearly explain monitoring and AI use to employees.
• Human oversight: keep accountable people involved in consequential decisions.
• Correction and challenge: give employees a meaningful route to question inaccurate data and decisions.
• Continuous review: check whether the technology remains necessary, fair and proportionate after implementation.

Employee Voice Is a Safeguard, Not an Obstacle

Employee consultation is sometimes viewed as something that slows technological adoption. The OECD reaches a different conclusion. Its 2025 analysis argues that worker consultation can support adoption while helping safeguard worker wellbeing.

Employees can identify harms that may be invisible to system designers. A monitoring metric may seem reasonable at management level but create pressure, gaming or unfairness when applied to real work. Employees can also explain which data feel intrusive and which forms of monitoring they consider legitimate.

Marchington and Wilkinson (2020) emphasise the importance of employee voice within the employment relationship. In a digital workplace, voice should extend to questions about data, algorithms and automated management.

Employees should not discover that they are being algorithmically managed only after a decision goes against them.

A Global HRM Perspective

Privacy and workplace monitoring also illustrate the difficulty of global HRM. Brewster et al. (2017) emphasise that multinational organisations operate across different legal, institutional and cultural environments.

The OECD evidence already shows different patterns of algorithmic-management adoption between the United States, Europe and Japan. Regulation also differs substantially. A multinational company may therefore have one global HR technology platform but face very different legal obligations and employee expectations regarding monitoring and data use in each location.

This creates a strategic choice. Should the organisation follow only the minimum legal requirement in each country, or should it establish a higher global ethical standard for employee data?

In my view, global HRM should aim for consistent principles—necessity, transparency, human oversight, fairness and employee voice—while adapting implementation to local legal and cultural contexts.

🎥 Recommended Video / Learning Resource

OECD AI-WIPS — AI in Work, Innovation, Productivity and Skills

The OECD AI-WIPS video collection explores how AI is affecting work, worker outcomes, productivity and policy, providing useful international context for the ethical questions discussed in this post.

▶ Watch the OECD AI-WIPS collection

My Reflection

Before examining this topic, I viewed employee data mainly as a useful management resource. If technology can provide more accurate information about attendance, productivity or performance, it seems logical that organisations would want to use it.

I now see that the ethical question begins much earlier than accuracy. HR should first ask whether the information should be collected at all, whether employees understand the purpose and whether the level of monitoring is proportionate to the organisational need.

The CIPD trust figures were particularly meaningful to me. The difference between 63% trusting AI to inform decisions and only 1% trusting it to make them suggests that employees may not be rejecting technology itself. Instead, they appear to be protecting the idea of human accountability.

The EU prohibition on workplace emotion recognition also changed my perspective. It demonstrates that technological capability should not automatically determine management practice. There must be boundaries around how deeply employers attempt to interpret and monitor employees.

For me, the future role of HR is therefore not simply to help organisations collect more workforce data. HR should help organisations decide which data deserve to be collected, how they should be used and where a human boundary should remain.

Employee trust is difficult to measure on a dashboard—but excessive monitoring can destroy it very quickly.

💬 Join the Discussion

How much workplace monitoring would you personally consider acceptable?

Would you accept AI analysing your work speed, emails, calls, location or wellbeing if your employer argued that it would improve productivity or safety? Where should HR draw the line? Share your view in the comments.

References

Brewster, C., Sparrow, P., Vernon, G. and Houldsworth, E. (2017) International Human Resource Management. 4th edn. London: CIPD.

CIPD (2025) Almost two thirds of people would trust AI to inform important work decisions, CIPD poll shows. 28 January. Available at: https://www.cipd.org/en/about/press-releases/almost-two-thirds-people-trust-ai-to-inform-important-work-decisions/ (Accessed: 12 August 2026).

Clegg, S., Courpasson, D. and Phillips, N. (2006) Power and Organizations. Newbury Park, CA: Pine Forge Press.

European Commission (2026) AI Act. Available at: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai (Accessed: 12 August 2026).

International Labour Organization (2025) Algorithmic management in the workplace. Available at: https://www.ilo.org/algorithmic-management-workplace (Accessed: 12 August 2026).

Marchington, M. and Wilkinson, A. (2020) Human Resource Management at Work. 7th edn. London: CIPD.

OECD (2023) ‘Ensuring trustworthy artificial intelligence in the workplace: Countries’ policy action’, in OECD Employment Outlook 2023: Artificial Intelligence and the Labour Market. Paris: OECD Publishing. Available at: OECD Employment Outlook 2023 (Accessed: 12 August 2026).

OECD (2025) How widespread is algorithmic management in workplaces? Paris: OECD Publishing. Available at: https://www.oecd.org/en/publications/how-widespread-is-algorithmic-management-in-workplaces_cda7a114-en/full-report.html (Accessed: 12 August 2026).