The Use Of Generative Artificial Intelligence in the Workplace: Legal Risks and Recommendations for Companies

The document defines "Generative Artificial Intelligence" ("GenAI") as artificial intelligence systems trained on large-scale datasets that are capable of producing content in various formats - such as text, images, video, audio, or software code - in response to user prompts. These tools may enhance efficiency across a wide range of business processes, including drafting e-mails, summarising texts, supporting research activities, and facilitating idea development. However, the widespread use of these technologies also entails various legal and institutional risks, particularly when deployed through third-party platforms.
 

A. The Risk of "Shadow Artificial Intelligence" ("Shadow AI")
 

The Authority defines "Shadow Artificial Intelligence" as the use of generative artificial intelligence tools by employees in business processes without the knowledge, approval, or oversight of the company for which they work.
 

Shadow AI use typically arises from employees seeking to accelerate their work or reduce routine task burdens. However, the transfer of meeting notes, internal correspondence, or information relating to corporate data to external generative artificial intelligence platforms may give rise to significant risks for companies.
 

The principal risks highlighted by the Authority are summarised below:
 

• Risks Relating to Decision Quality and Accuracy: When GenAI tools are used without being subject to any oversight or review process, they may produce misleading or inconsistent information.
   

• Risks Relating to the Protection of Intellectual Property and Trade Secrets: Sharing product information, business strategies, trade secrets, and other confidential information with GenAI tools may lead to such information being used in the training of artificial intelligence models, being disclosed, or losing corporate control over them, thereby increasing the risk of intellectual property infringement.
   

• Risks Relating to Corporate Reputation and Loss of Trust: The ability of GenAI tools to generate inaccurate yet convincing outputs that do not reflect reality may give rise to errors that are difficult to remedy within corporate processes and may result in reputational harm.
   

• Risks Relating to Information Security and Cybersecurity: The use of these tools outside corporate oversight may expand the attack surface to which companies are exposed by increasing the risks of malicious software and unauthorised access.
   

• Risks Relating to the Protection of Personal Data: The sharing of personal data with GenAI tools outside corporate oversight may give rise to risks of unlawful data processing and data breaches under the Personal Data Protection Law ("PDPL").
   

B. What Actions Can Companies Take?
 

The Authority notes that a blanket prohibition on the use of GenAI is neither practical nor advisable, as it may in fact encourage Shadow AI use. Companies are therefore recommended to establish a corporate policy grounded in guidance, balance, and awareness, rather than a prohibitive approach.
 

The principal actions that companies may consider in this regard are as follows:

• Establishing a Corporate Usage Policy: Companies should establish a policy that clearly sets out which GenAI tools may be used for which activities, what categories of data may be entered into these tools, and the limits of permitted use.
   

• Cautious Approach Regarding Sensitive Information and Personal Data: Care should be taken when sharing information that is sensitive from a corporate perspective, as well as personal data, with GenAI tools, and the privacy policies of the relevant tools should be carefully reviewed. It is recommended that anonymised, generalised, and abstract language be preferred over proper names or identifying expressions when interacting with these tools.
   

• Preventing Over-Reliance on Outputs (Automation Bias): Accepting system outputs as accurate without scrutiny and incorporating "hallucination" content - which may appear convincing yet does not reflect reality - into business processes should be avoided. Data obtained from GenAI tools should not be used as the basis for final corporate decisions; rather, they should be treated as supplementary inputs subject to human oversight and evaluation.
   

• Implementing Data Security and Access Control Measures: Employees should be required to use only tools that have been approved by the company and for which conditions of use have been established. In this context, network-level access restrictions, device-based limitations (such as restricting access to corporate devices only), and role-based access approaches may be considered.
   

• Raising Employee Awareness and Establishing Feedback Mechanisms: Policies established by the company should be regularly updated and communicated to employees. Training activities should be conducted to raise awareness of potential risks and matters requiring attention, and feedback mechanisms should be put in place to enable employees to share their experiences and the issues they encounter. This approach will contribute to increasing institutional awareness and identifying areas for improvement.
   

C. Conclusion
 

While generative artificial intelligence tools offer significant efficiency and speed advantages in business processes, the emergence of these technologies - particularly in the form of Shadow AI use - may give rise to serious legal and operational risks for companies in relation to data security, the protection of trade secrets, and the processing of personal data.
 

It is therefore important for companies to approach the use of generative artificial intelligence not merely as a technological efficiency tool, but also through the lens of corporate risk management and data protection. In this context, it is of considerable importance to establish clear and workable corporate policies, define limits on the sharing of sensitive data, strengthen access and oversight mechanisms, and raise employee awareness.

Preventive and guidance-oriented measures taken in this direction will contribute to the safe utilisation of the opportunities offered by generative artificial intelligence technologies and will significantly reduce the likelihood of companies encountering potential legal and reputational risks.