The hidden costs of AI-generated work go far beyond initial productivity gains, according to recent research from Stanford University and industry publications. While companies invest heavily in generative AI tools, studies show that up to 40% of employees encounter “workslop” — polished-looking but low-value AI outputs that require nearly two hours per incident to fix. For large employers, this inefficiency translates to millions in lost productivity each year as time is spent correcting, rewriting, or clarifying material.

Beyond finances, misuse of AI strains workplace relationships and trust. Many employees feel annoyed, confused, or offended by poor AI-generated drafts, and recipients often view senders as less creative or capable, harming future collaboration. Technical teams also face increased debugging, refactoring, and security risks, as AI-generated code can introduce hidden flaws and technical debt that destabilize systems over time. The true cost to organizations lies in shifting cognitive and correctional burdens downstream, often masking inefficiencies and eroding team dynamics, undermining the hoped-for benefits of AI productivity.