{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Canaries In The Wild","title":"Webinar: Context Bombs - stopping AI Attackers in their Tracks","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/231cadbd\"></iframe>","width":"100%","height":180,"duration":2921,"description":"What if you could turn prompt injection into a defence?\nTracebit's latest research shows how a single \"context bomb\" - a short string planted in a canary secret - can hijack an attacking AI agent's own safety guardrails and stop it in its tracks. Across 152 attack runs in an AWS cyber range, one context bomb cut full compromise from 36% of runs to 1%, and dropped admin privilege escalation from 57% to 5%.\nHear Tracebit's Alessandro Brucato (Security Researcher) and Sam Cox (Co-founder & CTO), alongside Gadi Evron (CEO & Founder, Knostic), walk through the research and take audience questions on what defensive prompt injection means for the future of AI-driven attacks.\n📄 Read the full research: https://agentic.tracebit.com/context-bombs/","thumbnail_url":"https://img.transistorcdn.com/WtwgUqqePwgZ1lV5VAiqLu-oeqLGM55mBORIXf5gSYI/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8zN2Ji/OTg0MDU1YmU4MmE0/YjUyMWY0MTQxN2Zk/ZWMzMC5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}