If 2025 was the Year of AI, 2026 is looking to be… well, also the Year of AI. While a lot of the language around AI and its use has already been established, we’re now seeing a raft of new terms associated with the widespread negative attitudes and consequences surrounding it. Terms such as nudify and AI psychosis describe societal problems arising from AI, while digital asbestos and ai;dr reflect, at the very least, a cautiousness around its overuse. It’s not all abject negativity, however, as we find new ways to manage the risks of a reliance on AI, with terms like human-in-the-loop.
nudify
Elon Musk’s Grok AI had to be reined in (again) earlier this year after users began asking it to nudify images of women posted on X, that is, to digitally manipulate the images so that they appear naked or semi-naked, without the subject’s knowledge or consent. While it is in theory possible to nudify such images without the use of AI, it is clear that the rise of AI has made such behaviour problematically accessible.
AI psychosis
AI psychosis is a proposed emotional condition in which an AI chatbot’s programmed validation and encouragement of a user’s presumptions (regardless of whether or not they are true) can cause symptoms of delusion and paranoia following heavy use. While it is not recognised as a clinical diagnosis, it has been supported by recent anecdotal reports in the media.
ai;dr
You might be more familiar with the predecessor to this term, tl;dr, short for ‘too long; didn’t read’, used to refer to content online which is perhaps unnecessarily long, or is at least longer than a user currently cares to read. Ai;dr, then, is a response to this in the face of the proliferation of AI slop throughout the internet. In this case, instead of the content being longer than is worth reading, a user might see reading AI generated text as a poor use of their time. Hence, ai;dr.
human-in-the-loop
As we adjust to the use of AI in more and more processes, we have also had to develop new systems for balancing the risks of AI with the efficiency it brings. Human-in-the-loop (HITL) describes systems which include a level of human supervision and decision-making in an otherwise largely automated system, allowing the risks of AI to be minimised while maintaining its benefits.
digital asbestos
Asbestos, as we all now know, can cause long term damage to those who are exposed to it. It’s dangerous, difficult to get rid of, and the real negative effects didn’t become apparent for a long time. Digital asbestos casts some modern technology as presenting similar risks, either to the person or to society. Included in this by some, of course, is AI. It’s suggested that AI is leading to long-term degradation of accuracy on the internet, and is difficult to remove once embedded.
