Why sniffing out AI use at work is easier said than done
Executive Take
Detection results are signals, not verdicts, so organizations without an explicit AI-use policy have effectively already decided that any and all AI use is permitted by default.
Executive Summary
Companies are weighing whether AI-detection tools like Pangram and GPTZero help or hurt workplace trust. Cases include Hachette pulling a novel after Pangram flagged 78% of text as AI-generated, and a Commonwealth Short Story Prize winner facing similar suspicion. Platforms like LinkedIn, Substack, and Qwoted now offer detection features, while Technical.ly and PwC avoid routine use, citing unreliability and false positives.
Why It Matters
HR and AI leaders need a stance now, because ambiguous detector results are already triggering firings, canceled contracts, and public retractions before policies catch up. Waiting to write an AI-use and disclosure policy means the decision gets made reactively, under scrutiny, instead of proactively.