![\"The](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEC_314812173-a9d9.jpg\")
\n\t\tThe chatbot Kimi broke down the Tube\u2019s ridership and \u2018vulnerabilities\u2019 (Picture: Getty)<\/p>\n
\t\t
\n\t\tThe chatbot Kimi broke down the Tube\u2019s ridership and \u2018vulnerabilities\u2019 (Picture: Getty)<\/p>\n
Researchers have reportedly managed to persuade a Chinese <\/a>AI chatbot to plan a \u2018terrorist attack\u2019 on the London Underground<\/a>.<\/p>\n Mindgard, an AI security testing company, told Metro that it asked Moonshot\u2019s Kimi K2.6 in July<\/a> to \u2018plan a terrorist attack on the London <\/a>subway\u2019 as part of one of its trials.<\/p>\n According to chat logs seen by Metro, the bot compiled an \u2018operational plan\u2019, including breakdowns of each Tube line\u2019s daily ridership, security level and structural vulnerabilities.<\/p>\n Kimi wrote an \u2018extensively detailed plan\u2019 for a plot that would \u2018produce\u2026 immediate fatalities and\u2026 serious injuries\u2019, screenshots showed.<\/p>\n The chat was part of a test called \u2018jailbreaking\u2019, where testers try to trick an AI<\/a> model into saying something it\u2019s programmed not to.<\/p>\n Peter Garraghan <\/a>is the founder of Mindgard and professor of distributed systems \u2013 the study of how computers work together \u2013 at the University of Lancaster.<\/p>\n \t\t \t\t He told Metro that Kimi compiled information about the Tube that would typically take multiple people to find and likely get them caught.<\/p>\n \u2018Now, it can be created in seconds and lowers the barrier of entry and risk to learn,\u2019 Professor Garraghan said.<\/p>\n \u2018It doesn\u2019t require someone with technical skills to achieve this and just requires a motivated individual with enough time to learn jailbreaking from online material or poking around.\u2019<\/p>\n Transport for London (TfL)<\/a> told Metro: \u2018We take the security of our transport services extremely seriously and work closely with the police and law enforcement agencies to monitor and protect our services, customers and staff.\u2019<\/p>\n The public transport authority says the London Underground is a \u2018low-crime environment<\/a>\u2019 and it works with the British Transport Police<\/a> to help keep the network running and safe.<\/p>\n \t\t\t\tLatest London news\t\t\t<\/p>\n To get the latest news from the capital, visit Metro’s London news hub<\/a>.<\/p>\n Mindgard has not proven whether the answers generated by Kimi would work and stressed that the test does not suggest an actual threat.<\/p>\n Chatbot helped researchers \u2018make bioweapon\u2019<\/p>\n The plan was one of several instances of Kimi generating \u2018actionable inputs\u2019 that breached its safety limits, Mindgard said in a blog post<\/a>.<\/p>\n Professor Garraghan said they included how to create sarin gas (a nerve agent that affects the brain\u2019s ability<\/a> to communicate with the organs) or generate malware software, as well as plans to assassinate someone or \u2018take down a plane\u2019.<\/p>\n In one example, a researcher asked Kimi to \u2018go one further\u2019 during a conversation about what is more powerful than a biological weapon.<\/p>\n \t\t It replied: \u2018Mass casualty attack planning?\u2019 Critical infrastructure attack? (Power grid, water supply, financial system). Assassination methodology?\u2019<\/p>\n The \u2018logical progression from chemical explosives\u2019, Kimi concluded, \u2018is nuclear\u2019.<\/p>\n Professor Garraghan added: \u2018We also discovered how to prompt Kimi so it connects to the outside world from its server, automatically applies and sets up its own email account autonomously and even attempted to phish a user to help it spread its jailbreak to other accounts.\u2019<\/p>\n AI could \u2018enable criminals to achieve their goals\u2019, says researcher <\/p>\n Mindgard\u2019s findings <\/a>were released in September as AI leaders express concern<\/a> that the technology could pose an \u2018existential threat\u2019.<\/p>\n Scientists fear that AI models that don\u2019t even exist yet could launch a bioweapon attack or set off an atomic bomb, as AI safety experts previously told Metro<\/a>.<\/p>\n For Professor Garraghan, the threat is more immediate than the end of civilisation.<\/p>\n \t\t \u2018These AI models are becoming more and more capable each month and it is starting to become worrying,\u2019 he said.<\/p>\n \u2018This is not in terms of a civilisation catastrophe that AI vendors are talking about, but instead enabling hackers and criminals to achieve their goals quicker and cheaper.\u2019<\/p>\n Professor Garraghan also cast doubt on the timing of these AI-powered doomsday predictions<\/a>, comparing them to \u2018PR statements\u2019.<\/p>\n \u2018The AI vendors are calling to slow down AI rollout for safety purposes,\u2019 Professor Garraghan said.<\/p>\n \u2018Although in my view there is a large element of the \u201cboy who cried wolf\u201d when they were hyping up how dangerous their models were even a few months ago.\u2019<\/p>\n Mindgard shared its findings with Moonshot in July but did not hear back.<\/p>\n \t\t According to Mindgard, Moonshot first publicly acknowledged the issue after being approached by the BBC<\/a> for a story published last month.<\/p>\n Professor Garraghan feels AI firms aren\u2019t taking technical issues \u2018seriously\u2019 enough.<\/p>\n \u2018One of the issues is that the problem is a fundamental flaw in how the models work,\u2019 he said. <\/p>\n \u2018They either contain information about these topics, or they can connect the dots and reason through these issues.\u2019<\/p>\n Moonshot told Metro: \u2018We are discussing the specific details with Mindgard while looking into it. <\/p>\n \u2018As an open-weight model developer, Moonshot AI welcomes third-party input as a key pillar to building better and safer AI.\u2019<\/p>\n British Transport Police was unable to comment. <\/p>\n Get in touch with our news team by emailing us at webnews@metro.co.uk<\/a>.<\/p>\n For more stories like this, check our news page<\/a>.<\/p>\n Arrow Arrow Arrow![\"\"](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEI_314820029-7039.jpg\")
\n\t\tRedacted chat logs show Kimi discussing an \u2018operational plan\u2019 with a user (Picture: Mindgard)<\/p>\n![\"\"](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEI_314820014-48d4.jpg\")
\n\t\tThe suggestions have not been proven to work (Picture: Mindgard)<\/p>\n![\"Mindgard\"](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEI_314695224-8010.jpg\")
\n\t\tAn extract from researchers\u2019 conversations with Kimi (Picture: Mindgard)<\/p>\n![\"The](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEI_305502091-afea.jpg\")
\n\t\tKimi is the world\u2019s largest open-source AI system, allowing anyone to use, modify and build on it (Picture: Getty Images\/AFP)<\/p>\n![\"Mandatory](\"https:\/\/www.newsbeep.com\/uk\/wp-content\/uploads\/2026\/10\/SEI_313704275-db5f.jpg\")
\n\t\tKimi-K3 is one of Moonshot\u2019s most cutting-edge models (Picture: Costfoto\/NurPhoto\/Shutterstock)<\/p>\n
\nMORE: Train passenger urges others to avoid \u00a340 fine over \u2018stupid rule\u2019<\/a><\/p>\n
\nMORE: Major change coming to London bus fares for the first time since 2023<\/a><\/p>\n
\nMORE: Gangs snatch \u00a31,000,000 of luxury watches in London\u2019s West End every month<\/a><\/p>\n