Uploaded April 2026 | Updated November 2026, 2 hours ago
Talk by Jason Kramer
socallinuxexpo.org/scale/23x/presentations/poison-once-compromise-many-how-model-reuse-amplifies-ai-vulnerabilities
AI models are rarely built from scratch. Through model reuse and transfer learning, organizations inherit risks they may never see. This session explores how backdoors, poisoning attacks, and evasion techniques survive across generations of models, exposing downstream systems to compromise.
Talk by Jason Kramer
socallinuxexpo.org/scale/23x/presentations/poison-once-compromise-many-how-model-reuse-amplifies-ai-vulnerabilities
AI models are rarely built from scratch. Through model reuse and transfer learning, organizations inherit risks they may never see. This session explores how backdoors, poisoning attacks, and evasion techniques survive across generations of models, exposing downstream systems to compromise.










