The State of Shadow AI 2026

distinct AI services in use at the median organization, counted across its network within the first weeks of visibility.

distinct AI services at one in four organizations. The upper quartile sees more than 120, and the highest observed is 175.

attempted data leaks to AI tools at the median organization: personal data, credentials, or financial data in a prompt, detected by the data protection layer. About 350 a week; a quarter of organizations see more than 600 a week.

organizations see at least one attempted data leak to an AI tool every day. Just over half see more than fifty a day.

of prompts to AI tools contain sensitive data, in the deployments where we have measured it at the prompt level.

of office professionals have used AI tools at work they believed were not permitted under company policy.

have shared work-related information with public AI tools: 34% customer data, 31% financial information or confidential documents.

of all employees use AI at work without IT oversight, and only 16% use it with oversight. Derived from Awareways: 75% of employees use AI for work, and 78% of that group does so without IT oversight.

of employees do not act on the AI rules they already know.

surge in daily AI usage among desk workers in six months.

of generative AI users access tools through personal accounts, bypassing enterprise controls.

breached organizations lack governance policies for managing AI or for spotting unapproved use, up from 63% a year earlier.

of AI applications in the workplace are visible to IT teams.

of employees know the rules around AI usage, yet the majority bypass them anyway.

of organizations have a formal, comprehensive AI policy, up from 28% a year earlier.

of breached organizations had a shadow AI incident, more than double the 20% a year earlier.

global average cost of a data breach, a record, up 12% in a year.

shadow AI incidents resulted in data loss or compromise; about one in five drew a regulatory fine.

of office professionals have put customer data into public AI tools, and 31% financial information or confidential documents.

increase in ChatGPT-themed phishing click rates in just two years (from 1.2% to 6.8%).

per user per month for Microsoft 365 Copilot. That's $1.8M a year for a 5,000-employee organization.

of organizations have achieved large-scale AI deployment. The rest risk massive shelfware.

SaaS overspend is expected through 2028 at organizations without centralized visibility into their SaaS portfolio.

of AI decision-makers report actual EBITDA lift from AI investments.

of new employees say AI access influences their choice of employer. Blocking AI means losing talent.

of workplace AI users are now on company-managed accounts, up from 25% a year earlier, while personal-account use fell from 78% to 47%. Giving people a sanctioned route moves them onto it.

to deploy Unseen Security and gain full Shadow AI visibility.