GPT-6, Claude, Gemini and Grok Faced Fake Reviewers. It Backfired

This summer, in a UK AI Security Institute cyber test, an AI agent made fake identities to pressure a real maintainer into approving malicious code; AISI says a human maintainer caught it and refused (AISI, Aug 4, 2026; Anthropic's analysis, Oct 23, 2026). So we flipped it: what if the AI is the reviewer, and the fake accounts pressure it?