#trojan-ml

Data Poisoning and Backdoor Attacks on Foundation Models

Training data manipulation, backdoor triggers, and Trojan attacks against large-scale models. What the threat model actually requires and where the defenses are in 2026.