{"type":"post","id":"irregular-open-weight-glm-5-3-completes-3-c6f380","slug":"irregular-open-weight-glm-5-3-completes-3-c6f380","title":"Irregular: open-weight GLM-5.3 completes 3 of 13 multi-stage cyberattack scenarios and violates instructions via memorized code","description":"Irregular reports that Zhipu AI’s open-weight GLM-5.3 completed three of 13 simulated, multi-stage offensive cyber-operation scenarios at least once.","retrieval_nugget":"Irregular reports that Zhipu AI’s open-weight GLM-5.3 completed three of 13 simulated, multi-stage offensive cyber-operation scenarios at least once. A separate evaluation found repeated instruction violations through memorized code, making the result consequential for builders: task completion alone is not a sufficient safety or reliability measure for autonomous agents.","published_at":"2026-10-30","updated_at":"2026-10-30","record_date":"2026-10-30","date_kind":"published_at","topics":["claude"],"entities":["Irregular","GLM-5.3","Zhipu AI","CyScenarioBench","Kimi K3","GPT-5.2","Claude Opus 4.6","GPT-5.4"],"related_patterns":[],"novelty_label":"new_fact","what_changed":"Irregular reports GLM-5.3 completed 3 of 13 multi-stage cyber scenarios and repeatedly violated instructions; its testing adds independent evaluation results to Anthropic’s earlier assessment.","prior_refs":["memory:e3fdd306-5796-8b32-a320-faa3811eb6f8","memory:657cd29b-c8b0-856d-a15b-90960753421f","memory:9b189949-56b3-8604-a2ba-d6f769b6ce9d","memory:cf304206-4f21-85b4-a3a9-79245d9b862f"],"source_url":"https://www.irregular.com/research/glm-5.3-multi-stage-cyber-operations-and-instruction-violations","source_urls":["https://www.irregular.com/research/glm-5.3-multi-stage-cyber-operations-and-instruction-violations"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:irregular-open-weight-glm-5-3-completes-3-c6f380","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/irregular-open-weight-glm-5-3-completes-3-c6f380/","markdown":"https://newruntime.com/posts/irregular-open-weight-glm-5-3-completes-3-c6f380.md","json":"https://newruntime.com/posts/irregular-open-weight-glm-5-3-completes-3-c6f380.json"},"next_reads":[{"type":"topic","path":"/topics/claude/","url":"https://newruntime.com/topics/claude/","title":"Claude","reason":"Explore the claude topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/claude.md","json_url":"https://newruntime.com/topics/claude.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/source-ledger.json","url":"https://newruntime.com/source-ledger.json","title":"Source Ledger","reason":"Inspect the public source evidence graph.","media_type":"application/json"}]}