Sharing confidential information is one of the activities that must be avoided at all costs, especially when working for a large company that has the culture of secret. But with the emergence of artificial intelligence, more and more employees use chatbots such as Chatgpt or Gemini in their daily life. According to a recent APEC survey, even more than a third would be to use it frequently.
However, some of them do so to facilitate their work, at the risk of involuntarily exposing private or confidential information on the Internet, with possible disastrous consequences such as data leaks. A drift called “Shadow Ai” (Ia Ghost) that worries the IT services of companies.
Risky clandestine use
We are talking about “Shadow AI” when we want to describe the “illegal” or not declared use of generative artificial intelligence. This can be used to generate text, photos or videos, but also to transform the data into a document or verify that an element is viable in a few minutes.
In general, Ais operates feeding the data they can receive in the context of a conversation with a user, to improve their future responses.
Anxious to use AI to support, facilitate? – His business work, 28% of French employees say they are overcome by the arrival of new technologies. Although more than 70% of them want to understand AI (78% in higher executives), according to a recent IPSO study, they often suffer an ignorance of possible uses within companies and the lack of internal training. IA is considered too often as harmless, while it can represent a long -term problem.
Develop useful tools for employees
However, not everything is negative in the “AI of the Shadow”. In a study recently published by Inria and Datacraft, a data scientists club, researchers show that this can also be “a strategic opportunity” capable of “transforming individual and diffuse practices into collective intelligence.”
Therefore, several recommendations are made to employers to stop the phenomenon of Shadow AI.
Without going so far as to prohibit the use of chatbot, the institute recommends communicating better in uses to gradually regulate the “informal” uses of a generative AI. This can go through collective bargaining about the regulation of AI uses (uses, executives, creation of spaces and tools used internally in a legal framework) or by creating training and/or workshops to define uses.
It is also required that companies make internal chatbot, especially so that the data does not leave the offices. In addition, and as the orange operator specifies, if an employee tends to use AI, it is essentially because the already accessible tools are not efficient enough or do not meet their needs.
Tip for good use
However, remember that, such as the pirated software, the unauthorized use of a generative AI in the workplace can be sanctioned. However, some individual security and security measures can be taken to safe uses:
- Do not enter confidential information, such as projects, client data or human resources, in chatgpt, geminis, claude, perplexity or co -pilot. Or anonymity data for analysis or processing.
- Do not install extensions or ia not validated applications in your workstation.
- Do not use AI to reuse protected content (text, photo …) and apply it as it is.
- See information to avoid any hallucination of AI (data invention).
- If you use both personnel and personnel, always favor the connection with your professional account. Tools such as Microsoft Copilot Anonymize the data if you are connected with a commercial account and do not use it to cause AI.
Source: BFM TV
