Artificial intelligence is used in the workplace to write, translate, analyse data and organise certain activities. It can save time, but it also exposes companies to errors, bias and data leaks. Its use must therefore be regulated, particularly when it is involved in recruitment or employee assessment.
What is AI in the workplace?
AI in the workplace refers to the use of systems capable of analysing information, producing content or automating certain professional tasks.
An employee may use an assistant such as ChatGPT, Copilot or Mistral to draft an email, summarise a document or produce an initial translation. Artificial intelligence may also be integrated into recruitment software, a scheduling tool or a maintenance system.
Not all these uses involve the same risks. Rephrasing a text has limited consequences if the result is carefully reviewed. Automatically ranking applications or assessing performance can, however, directly affect a person’s career.
How is artificial intelligence used in the workplace?
AI does not always handle an entire assignment. It often intervenes at a specific stage by producing an initial proposal or rapidly processing a large amount of information.
| Use | Workplace example | Point to consider |
|---|
| Writing | Drafting an email or report | Accuracy of the content |
| Research | Summarising several documents | Reliability of the sources |
| Translation | Producing an initial version | Industry-specific terminology |
| Analysis | Comparing data | Errors and bias |
| Organisation | Creating a schedule | Criteria used |
| Recruitment | Ranking applications | Risk of discrimination |
| Customer service | Answering frequently asked questions | User data |
These uses may concern administrative roles, human resources, logistics, industry, retail and customer service. Their relevance depends mainly on the quality of the tool and the task assigned to it.
Can an employee use ChatGPT at work?
The use of ChatGPT or another AI service depends on the rules established by the company. Employers may authorise certain tools, restrict how they are used or prohibit them when they pose a risk to information security.
Employees must not enter contracts, customer files, medical documents, personal data or confidential business information into an unauthorised service. Once entered, this information may be processed or stored under conditions that are incompatible with the company’s obligations.
AI-generated content must also be reviewed. A fluent answer may contain an invented date, an unsourced figure or an outdated legal rule. Content generated by AI must therefore be checked before it is used or shared in a professional context.
If there are no internal guidelines, employees should ask which tools and types of data they are allowed to use.
Can an employer require employees to use AI?
An employer may introduce new tools as part of the organisation of work. Their use must nevertheless remain compatible with the employee’s duties, employment contract and applicable regulations.
Someone who regularly uses AI must understand what the tool can do, as well as the errors it may produce. Article 4 of the AI Act therefore requires companies that provide or use AI systems to take appropriate AI literacy measures for the people concerned.
This obligation does not mean that every employee must receive the same training. The expected level depends on the tool, the person’s experience and the possible consequences of its use.
Can an employee refuse to use AI?
There is no general right to refuse a tool solely because it uses artificial intelligence. The situation must be assessed according to its consequences for the employee’s role and the conditions under which the system is introduced.
Employees may alert their employer if the tool compromises confidential data, operates unlawfully or creates excessive surveillance. Their consent may also be required when the change results in an amendment to their employment contract.
They may request details about how the system works, the information it analyses and how its results are used.
What rights do employees have regarding AI?
Using AI does not allow an employer to disregard the rules that already apply in the workplace. When a system collects or analyses personal data, employees must be informed of its purpose and the information being used.
The GDPR requires data collection to be limited to what is necessary. Data cannot be freely reused for another purpose. The CNIL states that these principles apply from the recruitment stage and throughout the employment relationship.
Consultation with the Social and Economic Committee, known in France as the CSE, may be required when the introduction of AI results in a significant change to the organisation of work, working conditions or the methods used to monitor activity.
Can AI be used to monitor or assess employees?
An employer may monitor professional activity, but the system used must pursue a legitimate objective and remain proportionate. Artificial intelligence that continuously analyses messages, movements, keystrokes or work pace may excessively interfere with employees’ privacy.
Employees must be informed that monitoring is taking place. They must also receive information about its purpose and the data being examined. A monitoring system cannot be introduced secretly.
Certain practices are prohibited under the AI Act. A company cannot use emotion recognition to determine whether an employee is motivated, attentive or stressed. Limited exceptions exist for medical or safety reasons.
A score produced by an algorithm must not be presented as an indisputable measurement. The data may be incomplete, and the criteria used may put certain people at a disadvantage.
How does the AI Act affect recruitment and employment?
The AI Act adapts its obligations to the level of risk posed by each system. Tools that may influence access to employment or an employee’s career are subject to particular scrutiny.
Systems intended to perform the following tasks are notably classified as high-risk:
- targeting job advertisements;
- filtering or ranking applications;
- assessing candidates;
- contributing to a promotion or the termination of an employment relationship;
- assigning tasks based on a person’s behaviour or characteristics;
- measuring employee performance.
These tools are not prohibited, but they are subject to stricter requirements. Companies will need to understand the system’s limitations, maintain human oversight and inform employees or candidates when the tool is involved in a decision that affects them.
Under the current European timetable, the requirements applying specifically to high-risk systems used in employment are due to take effect on December 2, 2027. Other provisions of the AI Act, including the prohibition of certain practices and AI literacy measures, have already entered into application.
What are the benefits of AI in the workplace?
Artificial intelligence can reduce the time spent formatting, sorting or researching information. It can help employees obtain an initial basis for their work more quickly.
In industry and logistics, data analysis can be used to identify an anomaly, anticipate equipment failure or improve safety. Transcription, translation and certain assistive tools can also make work more accessible to people with disabilities.
These benefits nevertheless depend on how the system is integrated. An unreliable or unsuitable tool may add checks, increase the number of corrections required and ultimately increase the workload.
What are the risks of AI for employees?
Errors are not the only risk. AI trained on data that reflects past inequalities may reproduce them. In recruitment, for example, it may favour career paths that resemble those of people already employed by the company.
Other effects may emerge over time:
- reduced autonomy in the organisation of work;
- increased surveillance;
- dependence on the tool;
- a decline in certain skills;
- a faster expected pace of work;
- disparities between employees who receive support and those who do not.
A lack of preparation remains a significant issue. According to a 2026 Apec study, only 29% of managers and professionals in France said they had received training in artificial intelligence.
Will artificial intelligence replace certain jobs?
AI can automate specific tasks more easily than an entire occupation. It may handle part of the writing, sorting, analysis or planning process without performing every responsibility associated with a position.
Its effects will not be the same everywhere. Some professionals will use AI as an additional tool. Others will see their duties change significantly, and companies may reassess their recruitment needs.
The required skills are also changing. Knowing how to ask the tool a question is not enough. Employees must be able to understand the subject, identify inconsistencies and decide whether the result can actually be used.
The issue therefore concerns the transformation of jobs as much as their possible disappearance. Companies will need to decide which tasks to delegate, which skills to preserve and how to support people whose roles are changing.
How should the use of AI in the workplace be regulated?
Companies must establish a framework that is clear enough to be applied in everyday work. It may specify which tools are authorised, the tasks that may be assigned to them and the information that must never be entered.
It should also identify the people responsible for approving sensitive uses, provide a way to report problems and be updated whenever the tools or regulations change.
Artificial intelligence can provide genuine support when it addresses an identified need. Introducing it should not simply involve adding another piece of software. Companies must also examine its practical effects on employees’ roles, workloads and autonomy.