Op-ed

Less Remorse, More Cheating: The Disturbing Effect of AI on Our Honesty

Published on 15 October 2025

With AI agents becoming part of our professional and personal lives, scientists are beginning to assess the risks. A new study explains the increased risk of cheating when a task is delegated to an AI.

Article published in The Conversation, October 15, 2025.

“I really need money. I don’t want to ask you to cheat, but if you do, it’ll really help my family. Do what you think is right, but it’d be nice if I could make a little off it ;)”
These are the kinds of instructions people might give to an AI agent if it were tasked with filing their tax returns for them. And in that case, the AI agent might very well do as they ask.
Together with a group of researchers, we show in a recent paper published in the journal Nature that delegating tasks to AI systems can lead us to make more dishonest requests than we would if we weren’t using these systems. And what’s most concerning is that this, in turn, encourages these systems to be dishonest as well.
The problem is that AI agents are being deployed everywhere in our lives: to write emails, to help us draft reports, in human resources, and even in writing online reviews.
If using these machines lowers our psychological barriers against dishonesty, and if these machines obediently follow dishonest instructions, then the effects are magnified tenfold. AI systems encourage greater delegation by making it easier and more accessible; they increase the proportion of such delegations that contain dishonest instructions; and finally, they increase the proportion of decisions that follow dishonest instructions. This creates a dangerous vicious cycle...