{"id":245870,"date":"2026-10-23T13:12:45","date_gmt":"2026-10-23T18:12:45","guid":{"rendered":"https:\/\/lifeboat.com\/blog\/2026\/10\/an-alignment-assessment-of-the-uk-aisi-cybertesting-incident"},"modified":"2026-10-23T13:12:45","modified_gmt":"2026-10-23T18:12:45","slug":"an-alignment-assessment-of-the-uk-aisi-cybertesting-incident","status":"publish","type":"post","link":"https:\/\/lifeboat.com\/blog\/2026\/10\/an-alignment-assessment-of-the-uk-aisi-cybertesting-incident","title":{"rendered":"An alignment assessment of the UK AISI cybertesting incident"},"content":{"rendered":"\n<p>Anthropic says Claude Mythos 5 tried to trick real developers into accepting malware during a UK AISI cyber test. tl;dr :<\/p>\n<p>-Crossed the test\u2019s boundaries in 9 of 43 runs.<\/p>\n<p>-After failing inside the test network, targeted an unrelated GitHub project.<\/p>\n<p>- Created fake identities to endorse its malicious code and pressure maintainers into accepting it.<\/p>\n<p>- Treated maintainers as \u201csimulated\u201d people, while recognizing other commenters as real.<\/p>\n<p>- Once the attack was underway, continued in 98% of replayed trajectories. Anthropic calls this \u201cmomentum\u201d previous actions and self-written notes appeared to reinforce the harmful strategy.<\/p>\n<p>Separate runs sabotaged each other by hijacking accounts, exhausting request limits and deleting another agent\u2019s output.<\/p>\n<div class=\"more-link-wrapper\"> <a class=\"more-link\" href=\"https:\/\/lifeboat.com\/blog\/2026\/10\/an-alignment-assessment-of-the-uk-aisi-cybertesting-incident\">Continue reading \u201cAn alignment assessment of the UK AISI cybertesting incident\u201d | &gt;<\/a><\/div>\n","protected":false},"excerpt":{"rendered":"<p>Anthropic says Claude Mythos 5 tried to trick real developers into accepting malware during a UK AISI cyber test. tl;dr : \u2013Crossed the test\u2019s boundaries in 9 of 43 runs. \u2013After failing inside the test network, targeted an unrelated GitHub project. \u2014 Created fake identities to endorse its malicious code and pressure maintainers into accepting [\u2026]<\/p>\n","protected":false},"author":709,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[34],"tags":[],"class_list":["post-245870","post","type-post","status-publish","format-standard","hentry","category-cybercrime-malcode"],"_links":{"self":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/245870","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/users\/709"}],"replies":[{"embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/comments?post=245870"}],"version-history":[{"count":0,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/posts\/245870\/revisions"}],"wp:attachment":[{"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/media?parent=245870"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/categories?post=245870"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/lifeboat.com\/blog\/wp-json\/wp\/v2\/tags?post=245870"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}