The use of LLM-based chatbots in organizations

Generative AI, most notably in the form of chatbots based on large language models (LLM), such as ChatGPT, Bard or Luminous, has the potential to completely transform the way organizations work. These AI technologies could enable organizations to optimize certain tasks by generating text and creating visual content, such as images or slides. In addition, AI technologies can increase productivity in daily work processes, as they make it possible for questions to be answered quickly, content to be created quickly and support to be provided in various professional areas. However, their use in organizations entails considerable legal risks, particularly in connection with data protection and intellectual property. The recent agreement reached by Member States on the AI Regulation and related discussions have given these concerns greater visibility and they have underscored the need for coherent regulation in this area.

Consequently, it is necessary to create comprehensive guidelines for the use of generative AI in organizations. It is essential that the use of these AI tools be regulated so as to strike a balance between harnessing their capabilities and mitigating the legal risks that come with their use in an organization. Not only must the protection of personal data be guaranteed, but the confidentiality of business and trade secrets must also be protected when such generative AI systems are used. In addition, steps must be taken to ensure that the IP rights of third parties are not infringed. It is therefore advisable to draw up guidelines for the use of LLM-based chatbots. Such guidelines can include the following measures:

• Specifying the conditions for the use of generative AI, e.g. defining the cases in which ChatGPT can be used
• Creating a company account
• Not allowing personal data to be entered
• Not allowing business and trade secrets to be entered
• Using the "opt-out" options for AI training and history (i.e. prompts and results are deleted)
• Not using unverified results, thereby ensuring that the rights of third parties are not infringed
• Raising the awareness of employees and providing regular training

The creation of guidelines for the use of LLM-based chatbots is an organization-specific task due to the different use cases, organizational needs and regulatory requirements. We stand ready to help you with the design and implementation of these guidelines, taking into account the individual requirements and the possible applications within the respective organization.