{"resourceId":"ny-its-advisory-2026-102-servicenow-ai-platform","versions":[{"version":"external-30eafcc53f555f3645cfb7ef7ec286c7a686e9cf8da99f7587da372d1fa0f271","resource":{"id":"ny-its-advisory-2026-102-servicenow-ai-platform","title":"ITS Advisory 2026-102: Multiple Vulnerabilities in ServiceNow's AI Platform Could Allow for Unauthorized Access","organization":"New York State Office of Information Technology Services","sector":"State government cybersecurity","geography":"New York, U.S.","publishedAt":"2026-09-25","publicationDate":"2026-09-25","eventDate":"2026-09-25","sourceName":"NY ITS","sourceLabel":"State government security advisory","sourceUrl":"https://its.ny.gov/2026-102","evidenceClass":"standards-guidance","outcomeClass":"cautionary","topics":["data-security","governance-procurement"],"finding":"NY ITS rated five ServiceNow AI Platform vulnerabilities (including unauthenticated improper access control, SQL injection and authorization bypass) as High risk for large and medium government entities and Medium for small ones, with no reported in-the-wild exploitation. It recommended applying vendor fixes after testing plus CIS safeguards: vulnerability management, monthly patching, least privilege, service-account inventory, penetration testing and network segmentation.","sledRelevance":"Direct state-government risk rating for ServiceNow tenants; useful as a government benchmark for how SLED security teams should prioritize ServiceNow advisories.","evidence":"Government advisory summarizing vendor CVEs with a government risk rating and control recommendations.","architectureImplications":"Recommends segmentation and least privilege around a SaaS platform increasingly connected to AI agents and integrations.","governanceImplications":"Shows a state CIO office treating ServiceNow AI Platform flaws as high-risk for government entities, supporting formal patch SLAs.","securityPrivacyImplications":"Unauthenticated data access and modification flaws affect instances holding employee, resident and case data.","caveats":"Published September 25, 2026, before this edition's window; included because today's October 2026 batch (KB3228994) is the next ServiceNow AI Platform advisory and this is the most recent government risk rating found. Not an independent technical analysis.","streamIds":["servicenow","state-government"],"roles":{"sales":"Sales framing: a state IT agency has formally rated ServiceNow AI Platform flaws high-risk for government. That framing will appear in customer security reviews. Account teams should be ready with patch-cadence and shared-responsibility explanations. Use it to support managed security or patch-assurance offers, not to downplay risk. Note it predates today's advisory and covers a different batch.","engineering":"Engineering framing: the advisory's CIS-based recommendations (patching, least privilege, service-account inventory, segmentation, penetration testing) form a practical checklist for ServiceNow tenants that integrate AI agents. Pre-sales should map how the customer's instance, integration users and MID servers meet each recommendation, and identify self-hosted or partner-managed exposure. The advisory lists Yokohama, Zurich and Australia patch thresholds; confirm current versions exceed them before addressing the newer October batch.","delivery":"Delivery framing: adopt the advisory's recommendations as recurring operating controls. Maintain a service-account inventory for integrations and agents, run patch reviews at least monthly, and include ServiceNow in penetration-testing scope. Track each monthly batch, including today's, through the same process so remediation is evidenced consistently. Limited applicability: it is a September advisory, included here as the government risk benchmark for today's batch rather than as new news."},"roleDetails":{"sales":{"customerProblem":"Government security teams need a defensible standard for prioritizing ServiceNow platform vulnerabilities.","stakeholders":["State CISO","Agency security officers","Platform owner"],"discoveryQuestions":["How do you rate and track SaaS platform advisories today?","Is ServiceNow in your penetration testing scope?"],"valueHypothesis":"Aligning to a state-issued rating can speed internal patch approvals.","potentialEngagement":"A control-gap review of a ServiceNow tenant against the advisory's recommendations.","unsupportedClaims":["That any New York instance was exploited","That the advisory covers the October 2026 CVEs"]},"engineering":{"fit":"Applies to SLED ServiceNow tenants on Yokohama, Zurich or Australia releases.","architectureIntegration":"Review integration and service accounts used by AI agents and MID servers.","prerequisites":["Release/patch inventory","Service-account inventory"],"constraints":["Advisory lists specific affected versions; later releases may differ"],"security":"Least privilege, segmentation and authenticated/unauthenticated testing of the instance.","proofOfValue":["Confirm versions exceed listed affected patches","Complete service-account review with owners and purposes"]},"delivery":{"work":"Operationalize advisory recommendations as recurring controls.","dependencies":["Security operations","Platform team","Penetration testing provider"],"ownership":"Agency CISO.","skillsAdoption":"Platform admins learn vulnerability-management workflows for SaaS.","governanceCheckpoints":["Monthly patch review","Quarterly service-account review"],"acceptanceCriteria":["Proposed: every integration account has a named owner and review date"],"risks":["Assuming SaaS hosting removes customer responsibility"]}},"retrievedAt":"2026-10-29T22:28:30-05:00","enrichedAt":"2026-10-29T22:33:17-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Not addressed.","procurementImplications":"Supports patch-SLA and security-testing clauses in ServiceNow partner contracts.","operatingModelImplications":"Positions ServiceNow patching as a recurring security operations task.","sourceVerification":{"openedUrl":"https://its.ny.gov/2026-102","referenceExcerpt":"Multiple vulnerabilities have been discovered in ServiceNow's AI Platform, the most severe of which could allow for unauthorized access.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}