Harmonic study: 64.5% of personal AI use at work goes undetected by companies
Nearly two-thirds of workers use personal AI accounts for job tasks, creating a massive visibility gap.
Get AI news that actually matters
One email a day. Zero fluff. Join 10,000+ professionals.
Harmonic's research on AI usage at work uncovers a critical visibility gap: 64.5% of all activity on personal and free AI accounts is actually for work purposes, meaning companies have no awareness or control over sensitive data being processed. Additionally, 45.6% of personal AI activity occurs on licensed plans that companies are already paying for, indicating employees are bypassing approved tools due to convenience.
Workers are not treating work and personal AI as separate — they use whichever tool is already open or easiest to access. Go-to-market teams have only 39% of their AI activity on approved tools, while operations teams have just 18%. Harmonic argues that measuring minutes (not queries) reveals deeper exposure: Claude sessions average 10m 12s vs ChatGPT's 5m 53s. Without SSO or streamlined authentication, employees default to personal accounts, leading to permanent IP loss when they leave. Companies must adopt universal single sign-on and match tools to team workflows rather than enforcing one-size-fits-all policies.
- 64.5% of personal AI account activity is for work tasks, completely invisible to employers
- 45.6% of personal AI use happens on company-paid licenses, showing widespread tool misuse
- Claude sessions average 10m 12s, nearly double ChatGPT's 5m 53s, indicating heavier data exposure
Why It Matters
Unmonitored AI use poses security and IP risks; companies need better authentication and team-specific tooling.