{
  "name": "Areebi Shadow AI Statistics",
  "description": "Verified statistics on unsanctioned AI use in the workplace, each linked to its primary source.",
  "licence": "CC BY 4.0",
  "attribution": "Areebi Research Team. \"Shadow AI Statistics 2026: 40+ Stats on Unsanctioned AI at Work.\" Areebi, June 2026. https://www.areebi.com/resources/research/shadow-ai-statistics",
  "source": "https://www.areebi.com/resources/research/shadow-ai-statistics",
  "lastUpdated": "10 June 2026",
  "statCount": 35,
  "stats": [
    {
      "id": "stat-orgs-data-exposed",
      "section": "hero",
      "value": "99%",
      "takeaway": "of organisations have sensitive data exposed to AI",
      "detail": "Varonis analysed nearly 10 billion files across 1,000 real-world environments and found essentially every organisation has sensitive data that AI tools could surface.",
      "sourceName": "Varonis, 2025 State of Data Security",
      "sourceUrl": "https://www.varonis.com/blog/state-of-data-security-report",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-orgs-data-exposed"
    },
    {
      "id": "stat-breach-share",
      "section": "hero",
      "value": "20%",
      "takeaway": "of data breaches involved shadow AI",
      "detail": "IBM's 2025 Cost of a Data Breach report identified shadow AI as a factor in one in five breaches, a brand-new category in the 2025 edition.",
      "sourceName": "IBM Cost of a Data Breach 2025",
      "sourceUrl": "https://www.ibm.com/reports/data-breach",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-breach-share"
    },
    {
      "id": "stat-breach-cost",
      "section": "hero",
      "value": "+$670K",
      "takeaway": "added to the average breach cost by shadow AI",
      "detail": "Breaches involving high levels of shadow AI cost roughly USD 670,000 more than those with low or no shadow AI (USD 4.63M versus USD 3.96M).",
      "sourceName": "IBM Cost of a Data Breach 2025",
      "sourceUrl": "https://www.ibm.com/reports/data-breach",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-breach-cost"
    },
    {
      "id": "stat-byoai",
      "section": "hero",
      "value": "78%",
      "takeaway": "of AI users bring their own AI to work",
      "detail": "Microsoft and LinkedIn's 2024 Work Trend Index (31,000 workers, 31 markets) found most AI users supply their own tools rather than waiting for an approved option.",
      "sourceName": "Microsoft 2024 Work Trend Index",
      "sourceUrl": "https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-byoai"
    },
    {
      "id": "stat-personal-accounts",
      "section": "hero",
      "value": "73%",
      "takeaway": "of workplace AI use runs through unsanctioned personal accounts",
      "detail": "Cyberhaven's telemetry found the large majority of corporate AI activity happens via personal accounts that bypass enterprise governance entirely.",
      "sourceName": "Cyberhaven, 2025 AI Adoption Report",
      "sourceUrl": "https://www.cyberhaven.com/blog/4-2-of-workers-have-pasted-company-data-into-chatgpt",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-personal-accounts"
    },
    {
      "id": "stat-no-governance",
      "section": "hero",
      "value": "63%",
      "takeaway": "of breached organisations had no AI governance policy",
      "detail": "IBM found nearly two-thirds of breached organisations lacked any policy governing AI use or detecting unauthorised use.",
      "sourceName": "IBM Cost of a Data Breach 2025",
      "sourceUrl": "https://www.ibm.com/reports/data-breach",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-no-governance"
    },
    {
      "id": "stat-chatgpt-exposures",
      "section": "hero",
      "value": "71%",
      "takeaway": "of sensitive-data exposures involved ChatGPT",
      "detail": "Across 22.4 million enterprise AI prompts, Harmonic Security attributed 71.2% of all sensitive-data exposures to ChatGPT alone.",
      "sourceName": "Harmonic Security, 2025",
      "sourceUrl": "https://www.harmonic.security/resources/what-22-million-enterprise-ai-prompts-reveal-about-shadow-ai-in-2025",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-exposures"
    },
    {
      "id": "stat-data-growth",
      "section": "hero",
      "value": "485%",
      "takeaway": "growth in corporate data sent to AI tools in one year",
      "detail": "Cyberhaven measured a 485% rise in the volume of corporate data flowing into AI tools between March 2023 and March 2024.",
      "sourceName": "Cyberhaven, 2025 AI Adoption Report",
      "sourceUrl": "https://www.cyberhaven.com/blog/4-2-of-workers-have-pasted-company-data-into-chatgpt",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-data-growth"
    },
    {
      "id": "stat-workers-use-ai",
      "section": "how-widespread",
      "value": "75%",
      "takeaway": "of knowledge workers already use generative AI at work",
      "detail": "Microsoft's 2024 Work Trend Index found three in four knowledge workers use generative AI at work, with 46% having started in the prior six months.",
      "sourceName": "Microsoft 2024 Work Trend Index",
      "sourceUrl": "https://www.microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now-comes-the-hard-part",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-workers-use-ai"
    },
    {
      "id": "stat-unverified-apps",
      "section": "how-widespread",
      "value": "98%",
      "takeaway": "of organisations have unverified apps including shadow AI",
      "detail": "Varonis found almost every organisation runs unsanctioned or unverified applications, increasing the risk of data exposure and breaches.",
      "sourceName": "Varonis, 2025 State of Data Security",
      "sourceUrl": "https://www.varonis.com/blog/state-of-data-security-report",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-unverified-apps"
    },
    {
      "id": "stat-leaders-suspect",
      "section": "how-widespread",
      "value": "69%",
      "takeaway": "of cybersecurity leaders have evidence or suspect employees use public GenAI at work",
      "detail": "A Gartner survey of cybersecurity leaders found more than two-thirds believe public generative AI is being used in their organisation, sanctioned or not.",
      "sourceName": "Gartner (via Infosecurity Magazine, Nov 2025)",
      "sourceUrl": "https://www.infosecurity-magazine.com/news/gartner-40-firms-hit-shadow-ai/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-leaders-suspect"
    },
    {
      "id": "stat-chatgpt-scale",
      "section": "how-widespread",
      "value": "800M+",
      "takeaway": "weekly active ChatGPT users by late 2025",
      "detail": "OpenAI reported ChatGPT surpassed 800 million weekly active users, the consumer tool most often reached for at work without sanction.",
      "sourceName": "OpenAI / TechCrunch",
      "sourceUrl": "https://techcrunch.com/2026/02/27/chatgpt-reaches-900m-weekly-active-users",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-scale"
    },
    {
      "id": "stat-tools-high-risk",
      "section": "how-widespread",
      "value": "71.7%",
      "takeaway": "of corporate AI tools present high or critical data-security risk",
      "detail": "Cyberhaven's 2025 assessment rated the majority of AI tools in corporate use as high or critical risk, with 83.8% of enterprise data going to AI flowing to risky tools.",
      "sourceName": "Cyberhaven, 2025 AI Adoption Report",
      "sourceUrl": "https://www.cyberhaven.com/press-releases/cyberhaven-report-majority-of-corporate-ai-tools-present-critical-data-security-risks",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-tools-high-risk"
    },
    {
      "id": "stat-sensitive-share",
      "section": "what-leaks",
      "value": "34.8%",
      "takeaway": "of corporate data put into AI tools is sensitive",
      "detail": "Cyberhaven found the sensitive share of data going into AI tools rose to 34.8%, up from 27.4% a year earlier and 10.7% two years earlier.",
      "sourceName": "Cyberhaven, 2025 AI Adoption Report",
      "sourceUrl": "https://www.cyberhaven.com/press-releases/cyberhaven-report-majority-of-corporate-ai-tools-present-critical-data-security-risks",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-sensitive-share"
    },
    {
      "id": "stat-prompts-sensitive",
      "section": "what-leaks",
      "value": "2.6%",
      "takeaway": "of enterprise AI prompts contained sensitive data",
      "detail": "Harmonic Security detected 579,113 sensitive instances across 22.4 million prompts in 2025 - a small share of an enormous and growing volume.",
      "sourceName": "Harmonic Security, 2025",
      "sourceUrl": "https://www.harmonic.security/resources/what-22-million-enterprise-ai-prompts-reveal-about-shadow-ai-in-2025",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-prompts-sensitive"
    },
    {
      "id": "stat-free-tier-exposures",
      "section": "what-leaks",
      "value": "16.9%",
      "takeaway": "of sensitive-data exposures flowed through free or personal accounts",
      "detail": "Harmonic found nearly one in six sensitive exposures came via free and personal-tier accounts - the usage that billing and licensing controls cannot see.",
      "sourceName": "Harmonic Security, 2025",
      "sourceUrl": "https://www.harmonic.security/resources/what-22-million-enterprise-ai-prompts-reveal-about-shadow-ai-in-2025",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-free-tier-exposures"
    },
    {
      "id": "stat-source-code",
      "section": "what-leaks",
      "value": "18.7%",
      "takeaway": "of sensitive data sent to AI is source code",
      "detail": "Source code was the single largest category of sensitive data put into AI tools in Cyberhaven's analysis, ahead of R&D material (17.1%).",
      "sourceName": "Cyberhaven, 2025 AI Adoption Report",
      "sourceUrl": "https://www.cyberhaven.com/press-releases/cyberhaven-report-majority-of-corporate-ai-tools-present-critical-data-security-risks",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-source-code"
    },
    {
      "id": "stat-no-access-controls",
      "section": "what-leaks",
      "value": "97%",
      "takeaway": "of organisations with an AI-related breach lacked proper AI access controls",
      "detail": "IBM found nearly all organisations that suffered an AI-related breach had no proper access controls governing AI - the control most predictive of exposure.",
      "sourceName": "IBM Cost of a Data Breach 2025",
      "sourceUrl": "https://www.ibm.com/reports/data-breach",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-no-access-controls"
    },
    {
      "id": "stat-chatgpt-confidential-paste",
      "section": "chatgpt-specific",
      "value": "11%",
      "takeaway": "of everything employees paste into ChatGPT is confidential",
      "detail": "Cyberhaven found that more than one in ten of all pastes into ChatGPT contained confidential company data.",
      "sourceName": "Cyberhaven Labs",
      "sourceUrl": "https://www.cyberhaven.com/blog/4-2-of-workers-have-pasted-company-data-into-chatgpt",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-confidential-paste"
    },
    {
      "id": "stat-chatgpt-egress-concentration",
      "section": "chatgpt-specific",
      "value": "0.9%",
      "takeaway": "of employees account for 80% of data egress into ChatGPT",
      "detail": "A tiny fraction of employees drive the overwhelming majority of confidential-data egress - a finding that shapes where detection and coaching should focus.",
      "sourceName": "Cyberhaven Labs",
      "sourceUrl": "https://www.cyberhaven.com/blog/4-2-of-workers-have-pasted-company-data-into-chatgpt",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-egress-concentration"
    },
    {
      "id": "stat-chatgpt-tried",
      "section": "chatgpt-specific",
      "value": "10.8%",
      "takeaway": "of knowledge workers have tried ChatGPT at work",
      "detail": "Cyberhaven's earlier measurement found roughly one in nine knowledge workers had used ChatGPT at work, with 8.6% having pasted data into it.",
      "sourceName": "Cyberhaven Labs",
      "sourceUrl": "https://www.cyberhaven.com/blog/4-2-of-workers-have-pasted-company-data-into-chatgpt",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-tried"
    },
    {
      "id": "stat-chatgpt-risk-share",
      "section": "chatgpt-specific",
      "value": "71.2%",
      "takeaway": "of sensitive-data risk concentrated in ChatGPT",
      "detail": "Across all AI tools Harmonic studied, ChatGPT alone accounted for 71.2% of the sensitive-data exposure risk.",
      "sourceName": "Harmonic Security, 2025",
      "sourceUrl": "https://www.harmonic.security/resources/what-22-million-enterprise-ai-prompts-reveal-about-shadow-ai-in-2025",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-chatgpt-risk-share"
    },
    {
      "id": "stat-copilot-delay",
      "section": "enterprise-response",
      "value": "40%",
      "takeaway": "of IT leaders delayed Microsoft 365 Copilot by 3+ months",
      "detail": "A Gartner survey of 132 IT leaders found data oversharing prompted 40% to delay their M365 Copilot rollout by three months or more.",
      "sourceName": "Gartner (via Computerworld)",
      "sourceUrl": "https://www.computerworld.com/article/3616459/microsoft-moves-to-stop-m365-copilot-from-oversharing-data.html",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-copilot-delay"
    },
    {
      "id": "stat-governance-resource",
      "section": "enterprise-response",
      "value": "64%",
      "takeaway": "said information governance and security risks consumed significant time during AI deployment",
      "detail": "In the same Gartner survey, nearly two-thirds reported that governance and security risk required significant time and resources to manage during rollouts.",
      "sourceName": "Gartner (via Computerworld)",
      "sourceUrl": "https://www.computerworld.com/article/3616459/microsoft-moves-to-stop-m365-copilot-from-oversharing-data.html",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-governance-resource"
    },
    {
      "id": "stat-gartner-2030",
      "section": "enterprise-response",
      "value": "40%",
      "takeaway": "of organisations will suffer shadow-AI security incidents by 2030",
      "detail": "Gartner predicts that by 2030 more than 40% of global organisations will experience security and compliance incidents caused by unauthorised AI tools.",
      "sourceName": "Gartner (via Infosecurity Magazine, Nov 2025)",
      "sourceUrl": "https://www.infosecurity-magazine.com/news/gartner-40-firms-hit-shadow-ai/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-gartner-2030"
    },
    {
      "id": "stat-technical-controls",
      "section": "enterprise-response",
      "value": "17%",
      "takeaway": "of companies have technical controls to prevent unauthorised AI uploads",
      "detail": "Analysis of IBM's findings indicates only about 17% of companies have technical controls to stop unauthorised data uploads to AI tools, leaving most reliant on policy and training alone.",
      "sourceName": "Kiteworks analysis of IBM 2025",
      "sourceUrl": "https://www.kiteworks.com/cybersecurity-risk-management/ibm-2025-data-breach-report-ai-risks/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-technical-controls"
    },
    {
      "id": "stat-deepseek-ban",
      "section": "australia-apac",
      "value": "Feb 2025",
      "takeaway": "Australia banned DeepSeek from government devices",
      "detail": "The Department of Home Affairs directed government entities to remove all DeepSeek products from Australian Government systems and devices over data-security and foreign-access concerns.",
      "sourceName": "The Register (Feb 2025)",
      "sourceUrl": "https://www.theregister.com/2025/02/05/australia_deepseek_ban/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-deepseek-ban"
    },
    {
      "id": "stat-apra-letter",
      "section": "australia-apac",
      "value": "Apr 2026",
      "takeaway": "APRA wrote to all regulated entities setting minimum AI-governance expectations",
      "detail": "APRA's letter to banks, insurers, and superannuation trustees found AI systems are too often deployed without inventory and that boards must maintain sufficient understanding of AI to provide effective oversight.",
      "sourceName": "APRA AI governance letter (analysis)",
      "sourceUrl": "https://insiconcyber.com/blog/apra-ai-governance-letter-2026",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-apra-letter"
    },
    {
      "id": "stat-china-tools",
      "section": "australia-apac",
      "value": "4.1%",
      "takeaway": "of enterprise AI traffic went to China-based tools",
      "detail": "Harmonic found 925,519 prompts (4.1% of AI usage) went to platforms such as DeepSeek and Kimi Moonshot - a data-residency concern central to Australia's DeepSeek ban.",
      "sourceName": "Harmonic Security, 2025",
      "sourceUrl": "https://www.harmonic.security/resources/what-22-million-enterprise-ai-prompts-reveal-about-shadow-ai-in-2025",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-china-tools"
    },
    {
      "id": "stat-samsung-leaks",
      "section": "incident-timeline",
      "value": "3 in 20 days",
      "takeaway": "Samsung suffered three source-code leaks within about 20 days (April 2023)",
      "detail": "Within roughly 20 days of allowing ChatGPT, Samsung engineers leaked semiconductor source code, defect-detection algorithms, and a confidential meeting transcript into the tool.",
      "sourceName": "TechCrunch (May 2023)",
      "sourceUrl": "https://techcrunch.com/2023/05/02/samsung-bans-use-of-generative-ai-tools-like-chatgpt-after-april-internal-data-leak/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-samsung-leaks"
    },
    {
      "id": "stat-samsung-ban",
      "section": "incident-timeline",
      "value": "1 May 2023",
      "takeaway": "Samsung banned generative AI on company devices and networks",
      "detail": "Samsung restricted ChatGPT, Bing, Bard, and similar tools on company-issued devices and internal networks, citing the difficulty of retrieving data once sent to external servers.",
      "sourceName": "TechCrunch (May 2023)",
      "sourceUrl": "https://techcrunch.com/2023/05/02/samsung-bans-use-of-generative-ai-tools-like-chatgpt-after-april-internal-data-leak/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-samsung-ban"
    },
    {
      "id": "stat-samsung-survey",
      "section": "incident-timeline",
      "value": "65%",
      "takeaway": "of Samsung staff said generative AI carries a security risk",
      "detail": "An internal Samsung survey conducted before the ban found about 65% of participants believed using generative AI tools carries a security risk.",
      "sourceName": "TechCrunch (May 2023)",
      "sourceUrl": "https://techcrunch.com/2023/05/02/samsung-bans-use-of-generative-ai-tools-like-chatgpt-after-april-internal-data-leak/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-samsung-survey"
    },
    {
      "id": "stat-palo-alto-protectai",
      "section": "market-context",
      "value": "~$700M",
      "takeaway": "Palo Alto Networks moved to acquire Protect AI (April 2025)",
      "detail": "Palo Alto Networks announced its intent to acquire AI-security firm Protect AI, with reporting placing the deal in the USD 650-700 million range, to anchor its Prisma AIRS platform.",
      "sourceName": "CNBC (April 2025)",
      "sourceUrl": "https://www.cnbc.com/2025/04/28/palo-alto-networks-to-buy-protect-ai-to-boost-artificial-intelligence-tools.html",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-palo-alto-protectai"
    },
    {
      "id": "stat-cisco-robust",
      "section": "market-context",
      "value": "Sep 2024",
      "takeaway": "Cisco acquired AI-security firm Robust Intelligence",
      "detail": "Cisco completed its acquisition of Robust Intelligence in September 2024 to secure AI applications across the enterprise; terms were not disclosed.",
      "sourceName": "SecurityWeek",
      "sourceUrl": "https://www.securityweek.com/cisco-to-acquire-ai-security-firm-robust-intelligence/",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-cisco-robust"
    },
    {
      "id": "stat-sentinelone-prompt",
      "section": "market-context",
      "value": "~$250M",
      "takeaway": "SentinelOne acquired Prompt Security (August 2025)",
      "detail": "SentinelOne agreed to acquire generative-AI security startup Prompt Security in a deal valued at approximately USD 250 million to advance its GenAI and agent-security strategy.",
      "sourceName": "SentinelOne / Dark Reading",
      "sourceUrl": "https://www.darkreading.com/endpoint-security/sentinelone-acquires-ai-startup-prompt-security",
      "permalink": "https://www.areebi.com/resources/research/shadow-ai-statistics#stat-sentinelone-prompt"
    }
  ]
}