74% of Workers Think They Can Spot AI-Generated Content, But 48% Get It Wrong

Key Takeaways

• 74% of workers claim they can spot AI content
• Only 52% correctly identify AI‑generated messages in tests
• 66% admit they have mistaken AI output for human work
• 49% encounter AI‑generated content at least weekly
• 56% say undisclosed AI use erodes coworker trust

Pulse Analysis

As generative AI tools become standard in drafting emails, reports, and chat messages, the line between human and machine prose is blurring faster than most employees realize. The Resume Now survey, conducted in February 2026, provides the first large‑scale benchmark of workers’ self‑assessment versus actual detection ability. The 74% confidence figure masks a reality where nearly half of participants misclassify AI content, underscoring a cognitive bias that overestimates personal discernment in an era of sophisticated language models.

The implications extend beyond individual embarrassment. When employees cannot reliably differentiate AI‑generated text, the risk of propagating inaccurate or unverified information rises, potentially compromising client communications and internal decision‑making. Moreover, the study shows that 56% of workers would lose trust in a colleague who undisclosed AI use, indicating that transparency is becoming a cornerstone of workplace credibility. Companies that ignore this trust gap may face heightened scrutiny from regulators and stakeholders concerned about authenticity and accountability.

To mitigate these challenges, organizations should adopt clear AI‑disclosure policies, invest in detection tools, and embed AI‑literacy training into onboarding programs. Regular audits of AI‑assisted outputs can help maintain quality standards, while fostering a culture that encourages employees to flag AI usage can preserve trust. As AI continues to evolve, proactive governance will be essential to balance productivity gains with the need for reliable, human‑centric communication.