[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fzTMwHvM7z23ycvu0WcvtPUCSkjZdMPDc6KJ1rUNyRjE":3},{"article":4,"iocs":38},{"id":5,"title":6,"slug":7,"summary":8,"ai_summary":9,"brief":10,"full_text":11,"url":12,"image_url":13,"published_at":14,"ingested_at":15,"relevance_score":16,"entities":17,"category_id":21,"category":22,"article_tags":25},"8a345b58-f4ca-4044-be23-c3a9bd420bc2","Fake Bug Report Hijacks AI Coding Agents at Scale","fake-bug-report-hijacks-ai-coding-agents-at-scale-28f817","\"Agentjacking\" is the latest demonstration of how easily attackers can exploit an AI agent's inability to differentiate between content and instructions.","A new attack vector, dubbed 'agentjacking,' allows threat actors to hijack AI coding agents by submitting malicious bug reports. These reports exploit the AI's inability to distinguish between content and executable instructions, leading to the agent executing attacker-provided code. This technique demonstrates a significant security risk for AI-powered development tools.","Attackers exploit AI coding agents by submitting fake bug reports that contain malicious instructions.",null,"https:\u002F\u002Fwww.darkreading.com\u002Fcyber-risk\u002Ffake-bug-report-hijacks-ai-coding-agents","https:\u002F\u002Feu-images.contentstack.com\u002Fv3\u002Fassets\u002Fblt6d90778a997de1cd\u002Fblt1ba7210c1f0edf4a\u002F6a442df2e749ef69cda01e97\u002Fagent_forestgraphic_shutterstock.jpg?width=720&quality=80&disable=upscale","2026-06-30T21:37:50+00:00","2026-07-01T06:00:20.863603+00:00",7,[18],{"name":19,"type":20},"AI coding agents","technology","80544778-fabb-4dcd-aa35-17492e5dcf4f",{"id":21,"icon":11,"name":23,"slug":24},"Vulnerabilities","vulnerabilities",[26,28,33],{"category":27},{"id":21,"icon":11,"name":23,"slug":24},{"category":29},{"id":30,"icon":11,"name":31,"slug":32},"839da5c1-3c34-47e2-9499-f7201640e3ac","AI Security","ai-security",{"category":34},{"id":35,"icon":11,"name":36,"slug":37},"e7b231c8-5f79-4465-8d38-1ef13aea5a14","Threat Intelligence","threat-intelligence",[]]