{"schema":"https://policywatcher.online/schemas/evidence-packet/v1","schemaVersion":"1.1.0","mappingVersion":"2026-07-29.1","changeId":"b2d5abd7-975d-4f45-8c59-14592b7007cb","screeningDate":"2026-10-23T11:40:12.896Z","publicationGate":"published","company":{"id":"2c5bac29-0643-4196-be90-4b5b2767b6ed","name":"OpenAI","slug":"openai","industry":"AI Provider"},"policy":{"id":"b4651c03-839a-4379-a4dd-08337ffac02a","name":"Usage Policies","type":"aup","jurisdiction":"Global","sourceUrl":"https://openai.com/policies/usage-policies/"},"sourceConfidence":{"state":"verified-retrieval","lastCheckedAt":"2026-10-23T11:40:12.884Z","retrievalChannel":"direct","dataStatus":"Available","publicSnapshotEvidence":true,"limitation":"Source confidence describes recorded retrieval and publication state. It does not rate the provider policy or certify source authenticity."},"snapshots":{"old":{"version":1,"sha256":"6ecf204b1e014f2db25b3fbeffcf3c4c2582bffdf094ac4400dc09929ae5e475","capturedAt":"2026-10-02T04:56:55.396Z"},"current":{"version":2,"sha256":"7d3127522218e9c7976f0af531dd23c0b7e240cf7f6c34ea1f3675d0533a0deb","capturedAt":"2026-10-23T11:40:12.884Z"}},"analysisLineage":{"status":"verified","requestSha256":"ff6c96f7ef4f32f8ad59fa8a1571b13af38d56d7691508668f17745c6484a02c","responseSha256":"cc54dcae3c6459d7f86fc01687c0bfa01f3057a512bda9678d5330c2f857b56c","model":"gemini-2.5-flash","promptVersion":"policy-analysis.v3-full-document","schemaVersion":"policy-analysis.schema.v2","inputCoverage":"complete","semanticCoverage":"not_measured","documents":{"old":{"sha256":"6ecf204b1e014f2db25b3fbeffcf3c4c2582bffdf094ac4400dc09929ae5e475","charLength":11184,"byteLength":11252,"rangeUnit":"utf16-code-unit","includedRanges":[[0,11184]],"excludedRanges":[]},"new":{"sha256":"7d3127522218e9c7976f0af531dd23c0b7e240cf7f6c34ea1f3675d0533a0deb","charLength":19533,"byteLength":19635,"rangeUnit":"utf16-code-unit","includedRanges":[[0,19533]],"excludedRanges":[]}},"boundary":"Hash verification links retained inputs and the application SDK request. It does not prove provider execution, semantic completeness or legal correctness."},"captureLineage":{"old":{"provenance":"unknown","sourceUrl":null,"requestedUrl":null,"finalUrl":null,"capturedAt":"2026-10-02T04:56:55.396Z","urlDisclosure":"credentials-query-fragment-redacted"},"current":{"provenance":"captured","sourceUrl":"https://openai.com/policies/usage-policies/","requestedUrl":"https://openai.com/policies/usage-policies/","finalUrl":"https://openai.com/policies/usage-policies/","capturedAt":"2026-10-23T11:40:12.884Z","urlDisclosure":"credentials-query-fragment-redacted"}},"assessment":{"summary":"New OpenAI policies restrict using outputs to train competing AI models, tighten rules on professional advice, and enhance minor protection.","overallRisk":"High","overallScore":7,"previousPublicChange":null,"scoreDelta":null,"direction":"baseline","reasons":[{"icon":"alert","textEn":"New restriction on using OpenAI outputs to train competing AI models.","textIt":"Nuova restrizione sull'uso degli output OpenAI per addestrare modelli AI concorrenti.","deltaScore":3,"evidenceQuote":"You must not use OpenAI models or their outputs to generate, prepare, select, or supply training material or signals for models that are intended as alternatives to or substitutes for OpenAI models or services.","evidenceSide":"new","relatedKpi":"kpiAiTrainingOptOut","anchorStatus":"verified"},{"icon":"warning","textEn":"Expanded prohibitions on AI tools providing tailored professional advice.","textIt":"Divieti estesi sugli strumenti AI che forniscono consulenza professionale personalizzata.","deltaScore":2,"evidenceQuote":"building or offering AI tools that provide others with tailored advice that requires a license, without appropriate involvement by a licensed professional. You also may not state or imply that the AI itself holds professional credentials or","evidenceSide":"new","relatedKpi":"kpiAutomatedDecision","anchorStatus":"verified"},{"icon":"warning","textEn":"Clarified 'high-stakes actions' and added 'law enforcement' to restricted uses.","textIt":"Chiarite 'azioni ad alto rischio' e aggiunta 'forze dell'ordine' agli usi ristretti.","deltaScore":2,"evidenceQuote":"automation of high-stakes actions in sensitive areas without human review. This includes actions, decisions, transactions, submissions, and other steps that materially affect a person’s rights, safety, access to essential services, or other","evidenceSide":"new","relatedKpi":"kpiAutomatedDecision","anchorStatus":"verified"}],"keyPoints":[{"textEn":"New policy prohibits using OpenAI outputs to train models that compete with OpenAI services.","textIt":"La nuova politica vieta l'uso degli output OpenAI per addestrare modelli concorrenti.","sentiment":"negative"},{"textEn":"Stricter rules apply to AI tools offering tailored professional advice, requiring licensed professional involvement.","textIt":"Regole più severe per gli strumenti AI che offrono consulenza professionale, richiedendo il coinvolgimento di professionisti autorizzati.","sentiment":"negative"},{"textEn":"Enhanced and more detailed safeguards are now in place to protect children and teens from various harms.","textIt":"Salvaguardie più dettagliate e rafforzate sono ora in vigore per proteggere minori da vari pericoli.","sentiment":"positive"},{"textEn":"The definition of 'high-stakes actions' in sensitive areas requiring human review has been clarified and expanded.","textIt":"La definizione di 'azioni ad alto rischio' in aree sensibili che richiedono revisione umana è stata chiarita ed estesa.","sentiment":"neutral"}],"regionImpacts":[{"region":"EU","perspective":"Enterprise","riskLevel":"High","impactAnalysisEn":"Enterprises face increased compliance burdens due to new restrictions on training competing models and stricter rules for professional advice. The expanded definition of high-stakes actions requires careful review to comply with the EU AI Act.","complianceNoteEn":"EU AI Act"},{"region":"EU","perspective":"Individual","riskLevel":"Low","impactAnalysisEn":"Individuals benefit from enhanced protections, especially regarding minors and clearer rules for high-stakes automated decisions. This aligns with the EU AI Act's focus on fundamental rights and safety.","complianceNoteEn":"EU AI Act"},{"region":"Global","perspective":"Enterprise","riskLevel":"High","impactAnalysisEn":"Global enterprises must conduct a thorough review of their AI development and deployment practices. The new restrictions on model training and professional advice, alongside clarified high-stakes automation, necessitate significant operational adjustments.","complianceNoteEn":null},{"region":"Global","perspective":"Individual","riskLevel":"Low","impactAnalysisEn":"Globally, individuals benefit from a more secure and responsible AI ecosystem, with enhanced protections for vulnerable groups like minors. The policy aims to prevent misuse and ensure ethical AI deployment worldwide.","complianceNoteEn":null},{"region":"US","perspective":"Enterprise","riskLevel":"High","impactAnalysisEn":"US businesses must adapt to new restrictions on using OpenAI outputs for model training, impacting AI development strategies. Stricter rules for professional advice and high-stakes automation require re-evaluation of existing AI applications.","complianceNoteEn":null},{"region":"US","perspective":"Individual","riskLevel":"Low","impactAnalysisEn":"Individuals gain from stronger safeguards for minors and clearer guidelines on AI-driven professional advice. These changes contribute to a safer digital environment, addressing growing concerns about AI misuse.","complianceNoteEn":null}],"explanationBoundary":"Score reasons and deltaScore values are stored AI-assisted screening outputs. Verified anchors confirm only that the quoted passage occurs in the named snapshot; they do not prove the interpretation."},"governance":{"boundary":"Mappings identify review relevance between recorded PolicyWatcher KPI fields and framework topics. They are not legal interpretations, conformity assessments, certifications or compliance verdicts.","mappings":[{"framework":{"id":"eu-ai-act","name":"Regulation (EU) 2024/1689 (EU AI Act)","shortName":"EU AI Act","referenceUrl":"https://eur-lex.europa.eu/eli/reg/2024/1689/oj","referenceVersion":"Official Journal text, 2024","reviewQuestion":"Which recorded policy statements may be relevant to transparency, automated decisions, data use and human oversight review?","kpiFields":["kpiAiTrainingOptOut","kpiAlgoTransparency","kpiAutomatedDecision","kpiAiBiasFairness"]},"status":"mapped","assessedCount":4,"mappedFieldCount":4,"evidence":[{"field":"kpiAiTrainingOptOut","label":"AI training opt-out","value":"Not Available"},{"field":"kpiAlgoTransparency","label":"Algorithmic transparency","value":"Opaque"},{"field":"kpiAutomatedDecision","label":"Automated decisions","value":"Partial"},{"field":"kpiAiBiasFairness","label":"AI bias and fairness","value":"Absent"}]},{"framework":{"id":"iso-42001","name":"ISO/IEC 42001:2023","shortName":"ISO/IEC 42001","referenceUrl":"https://www.iso.org/standard/42001","referenceVersion":"ISO/IEC 42001:2023 overview","reviewQuestion":"Which recorded policy statements may inform an AI management-system review of transparency, risk oversight and independent assurance?","kpiFields":["kpiAlgoTransparency","kpiAiBiasFairness","kpiIndependentAudit","kpiRegulatoryCompliance"]},"status":"mapped","assessedCount":4,"mappedFieldCount":4,"evidence":[{"field":"kpiAlgoTransparency","label":"Algorithmic transparency","value":"Opaque"},{"field":"kpiAiBiasFairness","label":"AI bias and fairness","value":"Absent"},{"field":"kpiIndependentAudit","label":"Independent audit","value":"Absent"},{"field":"kpiRegulatoryCompliance","label":"Regulatory compliance","value":"Comprehensive"}]},{"framework":{"id":"nist-ai-rmf","name":"NIST AI Risk Management Framework 1.0","shortName":"NIST AI RMF","referenceUrl":"https://www.nist.gov/itl/ai-risk-management-framework","referenceVersion":"AI RMF 1.0; NIST revision in progress, checked 2026-07-29","reviewQuestion":"Which recorded policy statements may support Govern, Map, Measure or Manage review questions?","kpiFields":["kpiAlgoTransparency","kpiAutomatedDecision","kpiAiBiasFairness","kpiContentModeration"]},"status":"mapped","assessedCount":4,"mappedFieldCount":4,"evidence":[{"field":"kpiAlgoTransparency","label":"Algorithmic transparency","value":"Opaque"},{"field":"kpiAutomatedDecision","label":"Automated decisions","value":"Partial"},{"field":"kpiAiBiasFairness","label":"AI bias and fairness","value":"Absent"},{"field":"kpiContentModeration","label":"Content moderation","value":"Opaque"}]},{"framework":{"id":"oecd-ai-principles","name":"OECD AI Principles","shortName":"OECD AI Principles","referenceUrl":"https://oecd.ai/en/ai-principles","referenceVersion":"OECD AI Principles, updated 2024","reviewQuestion":"Which recorded policy statements may be relevant to transparency, fairness, accountability and user agency review?","kpiFields":["kpiConsentMechanism","kpiAlgoTransparency","kpiAiBiasFairness","kpiIndependentAudit"]},"status":"mapped","assessedCount":4,"mappedFieldCount":4,"evidence":[{"field":"kpiConsentMechanism","label":"Consent mechanism","value":"Implicit"},{"field":"kpiAlgoTransparency","label":"Algorithmic transparency","value":"Opaque"},{"field":"kpiAiBiasFairness","label":"AI bias and fairness","value":"Absent"},{"field":"kpiIndependentAudit","label":"Independent audit","value":"Absent"}]}]},"humanReviewQuestions":["Does the original Usage Policies source still match the recorded public snapshot version 2?","Do the cited source passages support each displayed reason, KPI value and regional note?","Which advisory framework topics require specialist legal, risk or governance review for this use case?","Has a later public change superseded this packet before it is reused in a decision or publication?"],"methodologyUrl":"https://policywatcher.online/methodology/confidence","changeUrl":"https://policywatcher.online/change/b2d5abd7-975d-4f45-8c59-14592b7007cb","boundary":"This packet records PolicyWatcher evidence and AI-assisted screening for one public change. It is not legal advice, a compliance verdict, a certification, or proof that the external source remains unchanged.","contentDigest":"ac69e7c407d9eac1a3681769c001dab169ae978ef929218a3362d59c359ff904"}