2025影子AI状况报告：员工AI使用行为及其安全领导者启示

Table of Contents Finding 1 - 10 Shadow Al Apps Putting Your Data at Risk09Finding 2 - The Popularity Trap: High Adoption Doesn't Mean High Security11Finding 3 - OpenAl Accounts for 53% of All Shadow Al Usage Across Enterprises13Finding 4 - Shadow Al Isn't Temporary: Uncovering Months of Unsanctioned Al Usage15Finding 5 - Understaffed and Overexposed: 27% of Small Company Employees Use Shadow Al17 Executive Summary: The Shadow AI EraIs Here Security leaders face an unprecedented reality: Shadow Ai has infiltrated nearly every corner of theenterprise, creating massive blind spots that traditional security approaches cannot address.demand immediate action. l. The threat is real and it's massive. Weidentified the 10 riskiest Al applicationscurrently proliferating across our customerbase,withsecurityscoresso lowtheyshouldalarm any CiSO. Three applications (JivrusTechnologies, Happytalk, and Stability Al)receivedfailinggradesmeaningthattheylackfundamentalsecuritycontrols likeRBAC,MFA,and audit logging. These aren't just any tools,they're processing corporate data daily. 4.ShadowAIrunsdeeperthanmostrealizeThese tools do not disappear after the testingand experimentation ends. For example, someapps run unsanctioned for over 4oo days onaverage. In our study, we found CreativeX andSystem.com to have the longest standingaccess on average. Once embedded inworkflows formonths,these applicationsbecome nearly impossible to remove withoutdisrupting business operations and upsetting itsusers. Every day they persist, the security debtcompounds. 2. Mass adoption doesn't equal enterprisereadiness.The most widely adopted Al toolsaren't the most secure. CreativeX and Otter.aiboast thousands of users despite securityscores that should disqualify them fromenterprise use. Organizations are choosing Altools like they choose consumer apps: basedon features and convenience, not security 5.Smallerorganizationsfacedisproportionate risk. The smaller theorganization, the bigger the shadow Al problem.Companies with 11-50 employees show thehighest risk concentration: 27% of theirworkforce uses unsanctioned Al tools. Theseorganizations face the perfect storm: maximumAl adoption with minimum security resources tomanage it. 3. The OpenAI monopoly. OpenAl commands53% of all shadow Al usage across theorganizationsweassessed,processingdatafrom over10,000 enterprise users in our study.This unprecedented concentration means halfof all Al-related risk flows through a singleplatform. Any security incident, policy change,or service disruption at OpenAl couldsimultaneously impact the majority ofenterprise Al workflows. The bottom line: Shadow Al is here, running loose across enterprises andinvisible to traditional security tools. Smart security teamsare implementing shadow Al discovery and governancesolutions to turn this challenge into competitiveadvantage.Thepathforwardisclear:Aladoptionwon'tslowdownbecauseofsecurityconcerns.Securityteamsmust get ahead of shadow Al now or face mounting risksand compliance challenges later. Methodology Reco identified high-risk shadow Al applications through detailed analysis of anonymized, real-worldusage data collected across its customer base.This comprehensive assessment included: 1. Internal telemetry and Saas audit logs: Identifying unsanctioned Al apps actively usedby employees. 2. Evaluation across multiple security-relevant factors: : Total user count: Number of employees actively using each Al app.: Usage duration and frequency: Level and pattern of employee engagement with the app.:Registration type: Whether employees registered using corporate credentials orpersonalemailaccounts.: Authorization visibility: Assessment of whether apps integrated transparentlyvia standard corporate channels or operated covertly.: Security policy compliance: Alignment with essential enterprise security controls, such as SsO, dataretention policies, and encryption standards. 3. Correlation of policy violations and risk signals: : Data Loss Prevention (DLP) and shadow Al discovery alerts: Abnormal data flows or other suspicious activities linked to shadow Al app usage. 4.Detailed Security IndicatorAssessment: : Specific security indicators assessed included Encryption at Rest, Password Complexity, Auto.Renewal Subscription status, SsO Support, User Geo-Location Control, Content Security Policy (CSP)Audit Logs, Valid Certificate, Transport Security (HTTPS), 2FA Provisioning, Data Classification,Encryption Key Rotation, User Audit Logs, and Data Retention Policies. : Each indicator was classified with clear statuses (Pass, Warn, or Fail), contributing to a compositerisk score. Here's an example of how Reco assesses the risks of an AI app across 2O indicators: Bycorrelatingthese comprehensive indicators,Recoidentifiedthe shadow Alapplicationsposingtheindustry-wide perspective. The rigorous analysis approach ensured objective validation of each Al tool'ssecurity posture, providing a reliable basis for assessin