[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news:stories\u002F44:en:{}":3},{"story":4,"neighbours":39,"timeline":40,"suggestions":48,"indexable":70,"locale":71},{"id":5,"path":6,"slug":7,"headline":8,"summary":9,"summary_html":10,"why_it_matters":11,"section":12,"section_label":13,"quote":14,"author_name":14,"author_handle":14,"kind":15,"kind_label":16,"legal_status":14,"region":14,"importance":17,"source_name":18,"source_url":19,"source_published_at":20,"parent_id":14,"topics":21,"edition":30,"event":14,"created_at":33,"updated_at":33,"coverage":34},44,"\u002Fstories\u002F44-a-single-malicious-webpage-could-hijack-metas-muse-ai-agent","44-a-single-malicious-webpage-could-hijack-metas-muse-ai-agent","A single malicious webpage could hijack Meta's Muse AI agent","Security firm NebuSec showed in a lab that a webpage opened by Meta's Muse agent could pull out and alter a user's chat history and stored memories without any approval prompt, and by chaining a browser and a Linux kernel flaw, take over the agent's cloud machine including its credential store. NebuSec says Meta fixed the issues before publication; Meta hasn't commented.","\u003Cp>Security firm NebuSec showed in a lab that a webpage opened by Meta's Muse agent could pull out and alter a user's chat history and stored memories without any approval prompt, and by chaining a browser and a Linux kernel flaw, take over the agent's cloud machine including its credential store. NebuSec says Meta fixed the issues before publication; Meta hasn't commented.\u003C\u002Fp>","Before you let an AI agent browse the web with access to company accounts, give it its own low-privilege accounts and keep payment and admin credentials out of its reach; an approval layer alone isn't enough.","ai","AI & automation",null,"news","News",3,"NebuSec","https:\u002F\u002Fnebusec.ai\u002Fresearch\u002Fmeta-muse-credential-exposure\u002F?utm_source=news.codingdelta.com&utm_medium=web","2026-11-20T00:00:00+01:00",[22,26],{"slug":23,"name":24,"path":25,"items_count":14},"vulnerabilities","Vulnerabilities","\u002Ftopics\u002Fvulnerabilities",{"slug":27,"name":28,"path":29,"items_count":14},"ai-agents","AI agents","\u002Ftopics\u002Fai-agents",{"slug":31,"date":31,"path":32},"2026-11-22","\u002Feditions\u002F2026-11-22","2026-11-22T06:08:17+01:00",[35],{"source_name":36,"url":37,"title":14,"published_at":38},"The Hacker News","https:\u002F\u002Fthehackernews.com\u002F2026\u002F11\u002Fa-single-malicious-webpage-let.html?utm_source=news.codingdelta.com&utm_medium=web","2026-11-21T00:00:00+01:00",[],{"items":41,"edges":47},[42],{"id":5,"path":6,"slug":7,"headline":8,"summary":9,"summary_html":10,"why_it_matters":11,"section":12,"section_label":13,"quote":14,"author_name":14,"author_handle":14,"kind":15,"kind_label":16,"legal_status":14,"region":14,"importance":17,"source_name":18,"source_url":19,"source_published_at":20,"parent_id":14,"topics":43,"edition":46,"event":14,"created_at":33,"updated_at":33},[44,45],{"slug":23,"name":24,"path":25,"items_count":14},{"slug":27,"name":28,"path":29,"items_count":14},{"slug":31,"date":31,"path":32},[],[49],{"id":50,"path":51,"slug":52,"headline":53,"summary":54,"summary_html":55,"why_it_matters":56,"section":57,"section_label":58,"quote":14,"author_name":14,"author_handle":14,"kind":15,"kind_label":16,"legal_status":14,"region":14,"importance":59,"source_name":60,"source_url":61,"source_published_at":38,"parent_id":14,"topics":62,"edition":69,"event":14,"created_at":33,"updated_at":33},40,"\u002Fstories\u002F40-amazon-china-linked-hackers-ai-agents-found-and-exploited-a-zero-day-t","40-amazon-china-linked-hackers-ai-agents-found-and-exploited-a-zero-day-t","Amazon: China-linked hackers' AI agents found and exploited a zero-day themselves","Amazon's threat intelligence team says a suspected China-linked actor ran autonomous agents on the open-source Cairn pentest framework that found a path traversal zero-day in MONAI Label (0.8.5 and earlier), a medical imaging annotation tool, going from API mapping to file writes in about 11 minutes and then to root on university servers. Amazon spotted it through an unauthenticated Cairn instance; a fix is coming in the next MONAI Label release.","\u003Cp>Amazon's threat intelligence team says a suspected China-linked actor ran autonomous agents on the open-source Cairn pentest framework that found a path traversal zero-day in MONAI Label (0.8.5 and earlier), a medical imaging annotation tool, going from API mapping to file writes in about 11 minutes and then to root on university servers. Amazon spotted it through an unauthenticated Cairn instance; a fix is coming in the next MONAI Label release.\u003C\u002Fp>","AI agents cut the time between exposure and exploitation to minutes, so take any self-hosted tool that runs without authentication by default (AI, data or admin tools) off the internet or put it behind a login now.","security","Security",4,"AWS Security Blog","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Fsecurity\u002Fchinese-nexus-actors-cairn-ai-agents-found-a-zero-day-on-their-own\u002F?utm_source=news.codingdelta.com&utm_medium=web",[63,64,65],{"slug":27,"name":28,"path":29,"items_count":14},{"slug":23,"name":24,"path":25,"items_count":14},{"slug":66,"name":67,"path":68,"items_count":14},"open-source","Open source","\u002Ftopics\u002Fopen-source",{"slug":31,"date":31,"path":32},false,"en"]