Government guidelines for the ethical and responsible use of generative AI

The Central Digital and Data Office (CDDO) has recently published exhaustive guidelines for governing the use of generative AI tools. The primary objective is to ensure that government officials employ this technology in a lawful, ethical, and responsible manner, while retaining significant human oversight throughout the entire process.

This comprehensive document delineates ten guiding principles for utilising generative AI, which encompasses a broad spectrum of tools, including ChatGPT. These tools have the capability to independently produce written content, images, audio, video, or software code based on specific instructions.

David Knott, the CDDO’s chief technology officer for government, emphasised the dynamic and evolving nature of the guidance, reflecting the rapid advancement of generative AI. He articulated that the document is intended to be continuously updated as best practices emerge from experiences in utilising generative AI across government, industry, and society.

The key principles outlined in the guidelines underscore the importance of comprehending the limitations of generative AI, employing the technology in a lawful and ethical manner, implementing security measures, and maintaining significant human oversight at all stages of its utilisation.

The document also underscores the necessity of overseeing the entire generative AI lifecycle, utilising appropriate tools for specific tasks, upholding transparency and collaboration throughout the development process, and possessing the requisite skills and expertise to work with generative AI.

The release of this framework follows the issuance of preliminary introductory guidance for government employees on the use of generative AI six months prior. The introductory guidelines acknowledged the potential capabilities of tools like ChatGPT, while also advising against certain uses, such as creating authentic messages, government summaries, or policy documents.

Furthermore, the government is presently conducting trials of a chatbot powered by technology similar to ChatGPT, designed to aid users in locating content and answering questions on GOV.UK. The privacy notice for the GOV.UK Chat tool describes it as a “natural language interface” that provides human-like responses to user queries.

The comprehensive and detailed guidance represents a crucial step in ensuring that government officials handle and implement generative AI technology in a responsible and ethical manner. It stands as evidence of the government’s dedication to embracing new technologies while also prioritising accountability and ethical considerations in their usage.