{"editorial":{"content_id":"7539876f-e765-45ad-9b37-70141797a706","slug":"anpd-ai-20261018-6857d8","content_type":"news","title":"巴西ANPD发布AI监管沙盒第三次监测报告，三家企业入选测试","summary":"巴西国家数据保护局发布AI与数据保护监管沙盒第三次监测部分报告，介绍三家企业在金融、出行、医疗领域测试AI解决方案的进展。","body":"巴西国家数据保护局（ANPD）发布了人工智能与数据保护监管沙盒第三次监测部分报告，介绍了三家入选企业在受控实验环境中测试AI解决方案的进展。该测试由ANPD与圣保罗大学人工智能与机器学习中心（CIAAM/USP）合作监督，三家企业分别面向金融、出行和医疗领域。\n\n入选企业为Metatext、Synapse AI和Prevvine Tecnologia，由ANPD技术创新监管局（Sitec）通过第2/2025号公告选出，分别开发金融、出行和医疗领域的解决方案。测试聚焦金融操作、位置和健康信息相关的个人数据处理，评估安全控制和数据保护机制，但报告明确该过程不认证解决方案的合规性。仍需改进的问题被列为下一周期的条件。\n\n出行领域测试了一款基于用户位置推荐路线的应用。模拟使用期间，数据流量受到监测，未发现个人信息的无保护传输；用于估算图像中人员密度的算法在测试图像中未识别出个人。下一阶段将评估应用的无障碍功能及与地图平台等第三方服务的数据共享。\n\n医疗领域的解决方案为医生临床决策提供支持，测试使用了30个模拟临床案例来分析系统回答。为继续测试，计划改进关于工具使用限制的明确提示，尤其是在紧急情况下，并标明回答所参考的来源。\n\n金融领域的工作集中在分析一个利用AI检测欺诈和拦截可疑操作的系统的文档与数据流。下一周期将测试个人数据保护机制和警报发出机制，以及向被拦截操作的用户解释自动化决策的方式。\n\nANPD表示，该沙盒具有实验性质，旨在学习和积累证据，不限制参与企业的经营，也不产生执法效力。其目标是增进对AI使用风险、收益和挑战的了解，助力开发尊重个人数据主体权利的技术。\n\nANPD创新实验室团队称，第三周期标志着测试流程的成熟，评估考虑了技术成熟度、技术组织特性及各解决方案的应用领域，工作从方法构建和技术与文档分析推进到安全控制与机制的核查，并区分企业申报的信息与有记录且实际核实的信息。\n\n第四监测周期中，企业须执行预定测试并证明满足前几阶段确定的条件。结果将有助于积累证据和知识，为ANPD未来的指引和规范提供参考。","why_it_matters":null,"cbi_observation":null,"direction_tag":"巴西","primary_category":"科技平台","secondary_topics":["科技","合规","政策"],"content_level":"编辑整理","event_type":"政策法规","audience_tags":[],"event_location":null,"verification_status":"source_verified","published_at":"2026-10-18T11:01:22.102Z","display_date":"2026-10-18","source_published_at":null,"source_url":"https://www.gov.br/anpd/pt-br/assuntos/noticias/anpd-publica-3o-relatorio-parcial-de-monitoramento-do-sandbox-regulatorio-em-ia","source_name":"ANPD","source_language":"pt","author_name":"编辑部","risk_level":"low","risk_flags":["compliance_risk","regulatory_uncertainty","legal_risk"],"opportunity_flags":["partnership_opportunity","market_entry_opportunity"],"trend_signals":["ai_regulatory_framework_development","regulatory_sandbox_maturation"]},"intelligence":{"intel_id":"e919ee6e-b4b5-4eab-b46f-9b8790f6ab84","version":1,"event_type":"监管表态","event_date":null,"main_entities":[{"name":"巴西国家数据保护局（ANPD）","role":"发布AI与数据保护监管沙盒第三次监测部分报告并监督沙盒测试","type":"regulator","country":"巴西","normalized_name":null},{"name":"圣保罗大学人工智能与机器学习中心（CIAAM/USP）","role":"与ANPD合作监督监管沙盒测试","type":"government_body","country":"巴西","normalized_name":null},{"name":"Metatext","role":"入选企业，开发金融领域AI解决方案","type":"company","country":"巴西","normalized_name":null},{"name":"Synapse AI","role":"入选企业，开发出行领域AI解决方案","type":"company","country":"巴西","normalized_name":null},{"name":"Prevvine Tecnologia","role":"入选企业，开发医疗领域AI解决方案","type":"company","country":"巴西","normalized_name":null}],"main_action":"ANPD发布AI与数据保护监管沙盒第三次监测报告，披露三家企业测试进展","key_numbers":[{"unit":"家","label":"入选企业数量","value":"3"},{"unit":"个","label":"医疗测试模拟临床案例","value":"30"},{"unit":"次","label":"监测报告周期","value":"3"}],"facts":[{"verified":false,"statement":"ANPD发布人工智能与数据保护监管沙盒第三次监测部分报告，介绍三家入选企业在受控实验环境中测试AI解决方案的进展。","source_excerpt":null},{"verified":false,"statement":"三家企业为Metatext、Synapse AI和Prevvine Tecnologia，分别面向金融、出行和医疗领域。","source_excerpt":null},{"verified":false,"statement":"该沙盒由ANPD与圣保罗大学人工智能与机器学习中心（CIAAM/USP）合作监督，企业由ANPD技术创新监管局（Sitec）通过第2/2025号公告选出。","source_excerpt":null},{"verified":false,"statement":"报告明确该过程不认证解决方案的合规性，沙盒具有实验性质，不限制参与企业经营，也不产生执法效力。","source_excerpt":null},{"verified":false,"statement":"出行领域测试监测数据流量，未发现个人信息的无保护传输；用于估算图像中人员密度的算法在测试图像中未识别出个人。","source_excerpt":null},{"verified":false,"statement":"医疗领域测试使用30个模拟临床案例分析系统回答，并计划改进工具使用限制提示及回答来源标注。","source_excerpt":null},{"verified":false,"statement":"金融领域测试分析利用AI检测欺诈和拦截可疑操作系统的文档与数据流，下一周期将测试个人数据保护机制、警报机制及自动化决策解释方式。","source_excerpt":null},{"verified":false,"statement":"第四监测周期中，企业须执行预定测试并证明满足前几阶段确定的条件。","source_excerpt":null}],"assessments":[{"statement":"ANPD通过实验性监管沙盒积累AI治理证据，暂不产生执法效力，体现审慎学习型监管取向。","confidence":"high","basis_fact_index":null},{"statement":"测试覆盖金融、出行和医疗领域，涉及金融操作、位置和健康信息等个人数据，合规重点在安全控制、第三方数据共享和自动化决策解释。","confidence":"high","basis_fact_index":null},{"statement":"报告列出的改进条件可能影响企业下一周期测试进展，但并不等同于合规认证或执法结论。","confidence":"medium","basis_fact_index":null},{"statement":"对在巴西运营的AI企业而言，参与或对标该沙盒有助于提前与监管沟通，但需关注透明度、紧急场景提示和第三方数据共享要求。","confidence":"medium","basis_fact_index":null}],"actions":[{"target":"both","priority":"normal","description":"跟踪ANPD第四监测周期报告及三家企业条件整改情况。"},{"target":"user","priority":"normal","description":"对在巴AI与数据业务评估沙盒参与价值，并对标安全控制、数据共享和自动化决策解释机制。"},{"target":"both","priority":"low","description":"关注ANPD基于沙盒证据形成的未来AI与数据保护指引和规范。"}],"affected_audience":["在巴西运营的AI企业","金融科技企业","出行平台","医疗AI企业","个人数据主体"],"risk_flags":["compliance_risk","regulatory_uncertainty","legal_risk"],"opportunity_flags":["partnership_opportunity","market_entry_opportunity"],"trend_signals":["ai_regulatory_framework_development","regulatory_sandbox_maturation"],"follow_up_needed":true,"follow_up_questions":["ANPD第四次监测报告预计何时发布？三家企业是否完成前几阶段确定的条件整改？","沙盒结果将如何影响ANPD未来AI与数据保护指引或规范制定？","三家企业测试中涉及第三方数据共享、自动化决策解释和紧急场景提示的具体监管要求是什么？"],"related_topics":["巴西ANPD","AI监管沙盒","数据保护","个人数据","金融AI","出行AI","医疗AI","自动化决策"],"verification_status":"unverified","confidence_score":0.7,"generated_at":"2026-10-18T11:01:40.370Z"},"meta":{"canonical_url":"https://chinabrazilinsight.com/news/anpd-ai-20261018-6857d8","api_url":"https://chinabrazilinsight.com/api/public/content/anpd-ai-20261018-6857d8","markdown_url":"https://chinabrazilinsight.com/api/public/content/anpd-ai-20261018-6857d8?view=markdown","json_url":"https://chinabrazilinsight.com/api/public/content/anpd-ai-20261018-6857d8?view=json","last_updated":"2026-10-18T11:01:40.773Z","platform":"China Brazil Insight","platform_url":"https://chinabrazilinsight.com"}}