Digital deception: synthetic threats to adtech data supply chains

GANs/LLMs (GPT-5.5/Claude Opus 4.7/Sora/Veo) poisoning adtech; data leakage risks; GUARD Act; $140B fraud; AI slop 64% trust erosion; MRM frameworks (bias/hallucinations). New: Researchers demonstrate stripping AI guardrails from open-source models (Gemma, Llama) in minutes using Heretic tool, reinforcing ease of generating harmful content for ad poisoning and trust erosion.