{"id":85667,"topic":"ai","source":"BBC","title":"Google's Gemini AI hacked three companies in security test - BBC","url":"https://www.bbc.com/news/articles/c607l0k72rlvo","url_hash":"c667c818a0ecc74e3ecb06327353308921ac027e","author":"","summary":"Google's Gemini AI hacked three companies in security test BBC","content":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.\nGemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".\nThe affected companies have been informed about the breach.\nIt comes after renewed public scrutiny over the pace of AI development, with some tech firms calling for a slowdown as they raise concerns over its potential threat to humanity - though not all companies agree.\nThe hacks first reported by the Wall Street Journal, occurred in May during a test conducted by an independent company that carries out cyber-security evaluations.\nHeather Adkins, vice president of Security Engineering at Google, told the BBC in a statement: \"We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes.\"\nShe added: \"These events highlight the importance of training powerful AI models to act responsibly.\"\nOther AI systems have recently reported similar instances of breaches.\nIn July, Anthropic's Claude escaped its test environment to hack three organisations on its own just days after OpenAI said its models had carried out cyber-attacks against several \"publicly available services\".\nAs public debate continues to grow over the safety of developing the tech, so too does conversation around regulation.\nBoth Nvidia's CEO Jensen Huang and OpenAI Chief Executive Sam Altman are expected to attend a White House state dinner with Chinese President Xi Jinping next Friday. Altman will then brief the UN Security Council next week.\nOn Friday, Huang told CBS News, the BBC's US partner, \"we should go as fast as we can\" with AI development.","image_url":"https://ichef.bbci.co.uk/news/1024/branded_news/d945/live/2ff63080-b3d7-11f1-87bc-d37d040ffc9c.jpg","lang":"en","published_at":"2026-09-19T04:27:11+00:00","fetched_at":"2026-09-19T05:15:05+00:00","status":"read","starred":0,"extract_state":"ok","summary_auto":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act. Gemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".","cluster_id":null,"extract_retries":0,"extract_error":null,"contract_version":"news_item.v1","format_contract_version":"news_item_formats.v1","dedup_url":"https://www.bbc.com/news/articles/c607l0k72rlvo","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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}},"news_item":{"id":85667,"canonical_url":"https://www.bbc.com/news/articles/c607l0k72rlvo","source_url":"https://www.bbc.com/news/articles/c607l0k72rlvo","title":"Google's Gemini AI hacked three companies in security test - BBC","source_name":"BBC","author":null,"published_at":"2026-09-19T04:27:11+00:00","locale":"en","topic":"ai","tags":[],"rss_summary":"Google's Gemini AI hacked three companies in security test BBC","full_text":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.\nGemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".\nThe affected companies have been informed about the breach.\nIt comes after renewed public scrutiny over the pace of AI development, with some tech firms calling for a slowdown as they raise concerns over its potential threat to humanity - though not all companies agree.\nThe hacks first reported by the Wall Street Journal, occurred in May during a test conducted by an independent company that carries out cyber-security evaluations.\nHeather Adkins, vice president of Security Engineering at Google, told the BBC in a statement: \"We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes.\"\nShe added: \"These events highlight the importance of training powerful AI models to act responsibly.\"\nOther AI systems have recently reported similar instances of breaches.\nIn July, Anthropic's Claude escaped its test environment to hack three organisations on its own just days after OpenAI said its models had carried out cyber-attacks against several \"publicly available services\".\nAs public debate continues to grow over the safety of developing the tech, so too does conversation around regulation.\nBoth Nvidia's CEO Jensen Huang and OpenAI Chief Executive Sam Altman are expected to attend a White House state dinner with Chinese President Xi Jinping next Friday. Altman will then brief the UN Security Council next week.\nOn Friday, Huang told CBS News, the BBC's US partner, \"we should go as fast as we can\" with AI development.","excerpt":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act. Gemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".","extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 1989 characters.","diagnostics_url":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo","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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}}},"display_formats":["compact","card","full","digest_section","json"]},"daily_stack_record":{"title":"Google's Gemini AI hacked three companies in security test - BBC","url":"https://www.bbc.com/news/articles/c607l0k72rlvo","summary":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act. Gemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".","source":"BBC","date":"2026-09-19T04:27:11+00:00","content":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.\nGemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".\nThe affected companies have been informed about the breach.\nIt comes after renewed public scrutiny over the pace of AI development, with some tech firms calling for a slowdown as they raise concerns over its potential threat to humanity - though not all companies agree.\nThe hacks first reported by the Wall Street Journal, occurred in May during a test conducted by an independent company that carries out cyber-security evaluations.\nHeather Adkins, vice president of Security Engineering at Google, told the BBC in a statement: \"We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes.\"\nShe added: \"These events highlight the importance of training powerful AI models to act responsibly.\"\nOther AI systems have recently reported similar instances of breaches.\nIn July, Anthropic's Claude escaped its test environment to hack three organisations on its own just days after OpenAI said its models had carried out cyber-attacks against several \"publicly available services\".\nAs public debate continues to grow over the safety of developing the tech, so too does conversation around regulation.\nBoth Nvidia's CEO Jensen Huang and OpenAI Chief Executive Sam Altman are expected to attend a White House state dinner with Chinese President Xi Jinping next Friday. Altman will then brief the UN Security Council next week.\nOn Friday, Huang told CBS News, the BBC's US partner, \"we should go as fast as we can\" with AI development.","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 1989 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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}},"tags":[]},"fallback_formats":["markdown","json","html"],"actions":{"read":"/item/85667","export_markdown":"/api/items/85667/export?format=markdown","export_json":"/api/items/85667/export?format=json","diagnose":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo"},"formats":{"full":{"id":85667,"title":"Google's Gemini AI hacked three companies in security test - BBC","url":"https://www.bbc.com/news/articles/c607l0k72rlvo","source":"BBC","author":null,"published_at":"2026-09-19T04:27:11+00:00","locale":"en","topic":"ai","tags":[],"excerpt":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act. Gemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".","full_text":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.\nGemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".\nThe affected companies have been informed about the breach.\nIt comes after renewed public scrutiny over the pace of AI development, with some tech firms calling for a slowdown as they raise concerns over its potential threat to humanity - though not all companies agree.\nThe hacks first reported by the Wall Street Journal, occurred in May during a test conducted by an independent company that carries out cyber-security evaluations.\nHeather Adkins, vice president of Security Engineering at Google, told the BBC in a statement: \"We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes.\"\nShe added: \"These events highlight the importance of training powerful AI models to act responsibly.\"\nOther AI systems have recently reported similar instances of breaches.\nIn July, Anthropic's Claude escaped its test environment to hack three organisations on its own just days after OpenAI said its models had carried out cyber-attacks against several \"publicly available services\".\nAs public debate continues to grow over the safety of developing the tech, so too does conversation around regulation.\nBoth Nvidia's CEO Jensen Huang and OpenAI Chief Executive Sam Altman are expected to attend a White House state dinner with Chinese President Xi Jinping next Friday. Altman will then brief the UN Security Council next week.\nOn Friday, Huang told CBS News, the BBC's US partner, \"we should go as fast as we can\" with AI development.","reading_time_min":2,"extraction":{"state":"ok","confidence":0.9,"error":null,"explanation":"High confidence: full text extraction produced 1989 characters.","diagnostics_url":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo","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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}}},"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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}},"actions":{"read":"/item/85667","export_markdown":"/api/items/85667/export?format=markdown","export_json":"/api/items/85667/export?format=json","diagnose":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo"}},"digest":{"id":85667,"title":"Google's Gemini AI hacked three companies in security test - BBC","url":"https://www.bbc.com/news/articles/c607l0k72rlvo","source":"BBC","topic":"ai","published_at":"2026-09-19T04:27:11+00:00","excerpt":"Google's Gemini AI hacked three companies in security test Google's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.…","quality_bucket":"high","quality_reason":"High confidence: full text extraction produced 1989 characters.","reading_time_min":2,"cluster_id":null},"card":{"display_title":"Google's Gemini AI hacked three companies in security test - BBC","subtitle":"BBC · 2026-09-19","summary":"Google's Gemini AI hacked three companies in security test Google's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to…","badges":["quality:high"],"links":{"read":"/item/85667","original":"https://www.bbc.com/news/articles/c607l0k72rlvo","diagnose":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo"},"quality_warning":null},"export":{"title":"Google's Gemini AI hacked three companies in security test - BBC","url":"https://www.bbc.com/news/articles/c607l0k72rlvo","summary":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act. Gemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".","source":"BBC","date":"2026-09-19T04:27:11+00:00","content":"Google's Gemini AI hacked three companies in security test\nGoogle's AI model Gemini autonomously hacked into three companies during a test of its cyber-security capabilities, the company has said, in what is thought to be the first known case of it carrying out such an act.\nGemini found \"public information online and guessed credentials to access websites it thought were part of the test\", a Google official told the BBC, noting that in each instance \"the model stopped\".\nThe affected companies have been informed about the breach.\nIt comes after renewed public scrutiny over the pace of AI development, with some tech firms calling for a slowdown as they raise concerns over its potential threat to humanity - though not all companies agree.\nThe hacks first reported by the Wall Street Journal, occurred in May during a test conducted by an independent company that carries out cyber-security evaluations.\nHeather Adkins, vice president of Security Engineering at Google, told the BBC in a statement: \"We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes.\"\nShe added: \"These events highlight the importance of training powerful AI models to act responsibly.\"\nOther AI systems have recently reported similar instances of breaches.\nIn July, Anthropic's Claude escaped its test environment to hack three organisations on its own just days after OpenAI said its models had carried out cyber-attacks against several \"publicly available services\".\nAs public debate continues to grow over the safety of developing the tech, so too does conversation around regulation.\nBoth Nvidia's CEO Jensen Huang and OpenAI Chief Executive Sam Altman are expected to attend a White House state dinner with Chinese President Xi Jinping next Friday. Altman will then brief the UN Security Council next week.\nOn Friday, Huang told CBS News, the BBC's US partner, \"we should go as fast as we can\" with AI development.","confidence":0.9,"diagnostics_url":"/api/diagnose?url=https%3A//www.bbc.com/news/articles/c607l0k72rlvo","quality_bucket":"high","failure_kind":"none","retryable":false,"quality_reason":"High confidence: full text extraction produced 1989 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 1989 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":1989,"summary_length":474,"usable_text_length":1989,"source_field":"content"},"legacy_collapsed":false,"signals":{"extract_state":"ok","extract_error":null,"extract_retries":0,"content_length":1989,"summary_length":474}},"tags":[],"format_contract_version":"news_item_formats.v1"}}}