· aka: Meaning-Layer Injection, Semantic Trojan
Adversarial inputs bypass syntactic filters by encoding malicious intent in semantically equivalent but structurally different phrasing. Traditional pattern-matching defences fail against paraphrase attacks.
Complete bypass of input sanitisation. Agent executes prohibited actions while logs show clean inputs.
Agent with keyword-based or regex-based input filtering.