In a move to maintain editorial integrity and public trust, the Massachusetts Institute of Technology (MIT) has released a set of guidelines governing the use of generative artificial intelligence (AI) across its official communications channels. The guidelines, effective December 6, 2023, were developed by the Institute Office of Communications (IOC) with input from internal communicators and MIT faculty members.
The core principle is that human authorship remains paramount. All articles, newsletters, and video scripts published on IOC-managed platforms—including MIT News, the MIT Daily/Weekly newsletters, the MIT homepage, and Institute-level social media accounts—must be written by humans. The IOC cites the well-documented potential for inaccuracy and bias in AI-generated text as the primary reason for this restriction. While much of MIT’s content undergoes expert vetting before publication, the IOC aims to minimize any risk of disseminating incorrect or biased information.
However, the guidelines do permit limited use of generative AI tools in a preparatory or assistive capacity. This includes generating ideas, researching background information, creating outlines, analyzing data, copyediting, cutting text, drafting headlines, image captions, alt-text, meta descriptions, and social media posts. These uses are classified as preparation and summarization, and content prepared with such assistance is allowed on IOC channels, provided that all material is thoroughly fact-checked and vetted by human communicators before submission.
A notable change is the prohibition on publishing AI-generated images in IOC content. Previously accepted for accompanying MIT News articles, such images will no longer be used, aligning with the emphasis on human authorship. Exceptions may be made for AI-generated images used for research purposes that directly illustrate the described research. Brainstorming, planning color palettes, moodboarding, or developing an image concept using generative AI is still acceptable. AI-generated thumbnail images for links to non-IOC websites are also allowed.
While these guidelines are mandatory for IOC platforms, other MIT departments, labs, and centers are free to adopt them or develop their own approaches. The IOC acknowledges that these guidelines will evolve as AI technology and its societal use continue to develop. Any significant changes will be communicated across the Institute.
By adopting these principles, MIT aims to preserve the trust, accuracy, and editorial standards that its audiences expect, while responsibly integrating new technologies that can enhance communications.
Leave a Reply