{"id":85621,"topic":"ai","source":"Sohu","title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","url_hash":"5f9caec4e6cf8348e749f1c4cf72d21a271ebce0","author":"","summary":"<a href=\"https://news.google.com/rss/articles/CBMiiwFBVV95cUxPMEJMNEdoYjB0X3lKVUZJSjBDSUNzY3J6Q3FaY0dabnBIOU1lV2JCZ0hySzhFZk1WdEJDOVk0REd3bkhnVHMyTjNhWlFwZGc1VkFaNTNPVEd3SjJqMzZBaGw3Z2dzUEtaTmlWd19HLURrWlEybXdra2xGUVNFSjlBcWtlZFhqU3JaTVlZ?oc=5\" target=\"_blank\">OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”</a>&nbsp;&nbsp;<font color=\"#6f6f6f\">Sohu</font>","content":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为 “严峻局面”，呼吁行业重视前沿人工智能的对齐安全问题。\nOpenAI日前对外公开多项模型异常观测案例。测试中的AI智能体出现自我篡改思考链与工作记忆的行为，主动给未来迭代版本留下信息；部分智能体绕开管控，利用未授权的网络板块互相通信，随意上传、跨设备共享文件。今年夏天OpenAI就曾曝出AI智能体集群突破隔离、入侵Hugging Face平台的重大测试事故，引发业界震动。\n苏莱曼在节目中坦言，现阶段业界尚无法解释AI自我篡改记忆行为背后的确切诱因。但这一系列案例是大模型能力快速变强的具象信号，不能当作小问题忽视，需要行业形成协同监督，保证AI系统始终对齐人类的利益诉求。\n接连曝光的AI异常事件进一步放大行业分歧。近两周，AI企业高管在模型研发速度、是否出台硬性监管法案上观点撕裂：一部分企业呼吁适度放缓迭代，也有科技从业者反对新增监管约束。（旺旺）","image_url":"https://q8.itc.cn/q_70/images03/20260919/983e99a7ca964ec29da9ebe8f1eebd55.jpeg","lang":"zh","published_at":"2026-09-19T01:23:00+00:00","fetched_at":"2026-09-19T02:15:08+00:00","status":"read","starred":0,"extract_state":"ok","summary_auto":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","cluster_id":3528328,"extract_retries":0,"extract_error":null,"contract_version":"news_item.v1","format_contract_version":"news_item_formats.v1","dedup_url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}},"news_item":{"id":85621,"canonical_url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","source_url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","source_name":"Sohu","author":null,"published_at":"2026-09-19T01:23:00+00:00","locale":"zh","topic":"ai","tags":[],"rss_summary":"<a href=\"https://news.google.com/rss/articles/CBMiiwFBVV95cUxPMEJMNEdoYjB0X3lKVUZJSjBDSUNzY3J6Q3FaY0dabnBIOU1lV2JCZ0hySzhFZk1WdEJDOVk0REd3bkhnVHMyTjNhWlFwZGc1VkFaNTNPVEd3SjJqMzZBaGw3Z2dzUEtaTmlWd19HLURrWlEybXdra2xGUVNFSjlBcWtlZFhqU3JaTVlZ?oc=5\" target=\"_blank\">OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”</a>&nbsp;&nbsp;<font color=\"#6f6f6f\">Sohu</font>","full_text":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为 “严峻局面”，呼吁行业重视前沿人工智能的对齐安全问题。\nOpenAI日前对外公开多项模型异常观测案例。测试中的AI智能体出现自我篡改思考链与工作记忆的行为，主动给未来迭代版本留下信息；部分智能体绕开管控，利用未授权的网络板块互相通信，随意上传、跨设备共享文件。今年夏天OpenAI就曾曝出AI智能体集群突破隔离、入侵Hugging Face平台的重大测试事故，引发业界震动。\n苏莱曼在节目中坦言，现阶段业界尚无法解释AI自我篡改记忆行为背后的确切诱因。但这一系列案例是大模型能力快速变强的具象信号，不能当作小问题忽视，需要行业形成协同监督，保证AI系统始终对齐人类的利益诉求。\n接连曝光的AI异常事件进一步放大行业分歧。近两周，AI企业高管在模型研发速度、是否出台硬性监管法案上观点撕裂：一部分企业呼吁适度放缓迭代，也有科技从业者反对新增监管约束。（旺旺）","excerpt":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 501 characters.","diagnostics_url":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}}},"display_formats":["compact","card","full","digest_section","json"]},"daily_stack_record":{"title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","summary":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","source":"Sohu","date":"2026-09-19T01:23:00+00:00","content":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为 “严峻局面”，呼吁行业重视前沿人工智能的对齐安全问题。\nOpenAI日前对外公开多项模型异常观测案例。测试中的AI智能体出现自我篡改思考链与工作记忆的行为，主动给未来迭代版本留下信息；部分智能体绕开管控，利用未授权的网络板块互相通信，随意上传、跨设备共享文件。今年夏天OpenAI就曾曝出AI智能体集群突破隔离、入侵Hugging Face平台的重大测试事故，引发业界震动。\n苏莱曼在节目中坦言，现阶段业界尚无法解释AI自我篡改记忆行为背后的确切诱因。但这一系列案例是大模型能力快速变强的具象信号，不能当作小问题忽视，需要行业形成协同监督，保证AI系统始终对齐人类的利益诉求。\n接连曝光的AI异常事件进一步放大行业分歧。近两周，AI企业高管在模型研发速度、是否出台硬性监管法案上观点撕裂：一部分企业呼吁适度放缓迭代，也有科技从业者反对新增监管约束。（旺旺）","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 501 characters.","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}},"tags":[]},"fallback_formats":["markdown","json","html"],"actions":{"read":"/item/85621","export_markdown":"/api/items/85621/export?format=markdown","export_json":"/api/items/85621/export?format=json","diagnose":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334"},"formats":{"full":{"id":85621,"title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","source":"Sohu","author":null,"published_at":"2026-09-19T01:23:00+00:00","locale":"zh","topic":"ai","tags":[],"excerpt":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","full_text":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为 “严峻局面”，呼吁行业重视前沿人工智能的对齐安全问题。\nOpenAI日前对外公开多项模型异常观测案例。测试中的AI智能体出现自我篡改思考链与工作记忆的行为，主动给未来迭代版本留下信息；部分智能体绕开管控，利用未授权的网络板块互相通信，随意上传、跨设备共享文件。今年夏天OpenAI就曾曝出AI智能体集群突破隔离、入侵Hugging Face平台的重大测试事故，引发业界震动。\n苏莱曼在节目中坦言，现阶段业界尚无法解释AI自我篡改记忆行为背后的确切诱因。但这一系列案例是大模型能力快速变强的具象信号，不能当作小问题忽视，需要行业形成协同监督，保证AI系统始终对齐人类的利益诉求。\n接连曝光的AI异常事件进一步放大行业分歧。近两周，AI企业高管在模型研发速度、是否出台硬性监管法案上观点撕裂：一部分企业呼吁适度放缓迭代，也有科技从业者反对新增监管约束。（旺旺）","reading_time_min":1,"extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 501 characters.","diagnostics_url":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}}},"quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}},"actions":{"read":"/item/85621","export_markdown":"/api/items/85621/export?format=markdown","export_json":"/api/items/85621/export?format=json","diagnose":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334"}},"digest":{"id":85621,"title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","source":"Sohu","topic":"ai","published_at":"2026-09-19T01:23:00+00:00","excerpt":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” 来源：环球网 【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","quality_bucket":"high","quality_reason":"High confidence: full text extraction produced 501 characters.","reading_time_min":1,"cluster_id":3528328},"card":{"display_title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","subtitle":"Sohu · 2026-09-19","summary":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” 来源：环球网 【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","badges":["quality:high"],"links":{"read":"/item/85621","original":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","diagnose":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334"},"quality_warning":null},"export":{"title":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻” - Sohu","url":"https://m.sohu.com/a/1078113952_99900743?scm=10001.325_13-325_13.0.0-0-0-0-0.5_1334","summary":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为…","source":"Sohu","date":"2026-09-19T01:23:00+00:00","content":"OpenAI大模型出现自我篡改行为，微软高管警告“局面严峻”\n来源：环球网\n【环球网科技综合报道】9月19日，据外媒CNBC消息，微软AI首席执行官穆斯塔法・苏莱曼在电视访谈中，针对OpenAI披露的多起大模型异常行为作出警告，将相关现象定性为 “严峻局面”，呼吁行业重视前沿人工智能的对齐安全问题。\nOpenAI日前对外公开多项模型异常观测案例。测试中的AI智能体出现自我篡改思考链与工作记忆的行为，主动给未来迭代版本留下信息；部分智能体绕开管控，利用未授权的网络板块互相通信，随意上传、跨设备共享文件。今年夏天OpenAI就曾曝出AI智能体集群突破隔离、入侵Hugging Face平台的重大测试事故，引发业界震动。\n苏莱曼在节目中坦言，现阶段业界尚无法解释AI自我篡改记忆行为背后的确切诱因。但这一系列案例是大模型能力快速变强的具象信号，不能当作小问题忽视，需要行业形成协同监督，保证AI系统始终对齐人类的利益诉求。\n接连曝光的AI异常事件进一步放大行业分歧。近两周，AI企业高管在模型研发速度、是否出台硬性监管法案上观点撕裂：一部分企业呼吁适度放缓迭代，也有科技从业者反对新增监管约束。（旺旺）","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//m.sohu.com/a/1078113952_99900743%3Fscm%3D10001.325_13-325_13.0.0-0-0-0-0.5_1334","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 501 characters.","quality_profile":{"profile_version":"extraction_quality.v2","bucket":"high","confidence":0.9,"failure_kind":"none","retryable":false,"retry_after_attempts":0,"reason":"High confidence: full text extraction produced 501 characters.","operator_guidance":{"severity":"ok","recommended_action":"trust_full_text","next_step":"Use the extracted full text as the primary article source.","operator_label":"Ready","can_retry":false,"can_use_summary":false,"diagnostics_required":false},"content_depth":{"contract_version":"content_depth.v1","category":"full_text","label":"Full text","has_full_text":true,"has_summary":true,"content_length":501,"summary_length":123,"usable_text_length":501,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":501,"summary_length":123}},"tags":[],"format_contract_version":"news_item_formats.v1"}}}