Viewing profile — spkavanagh6

A researcher demonstrates a novel LLM manipulation technique called 'Runtime Alignment Context Injection' (RACI) against Claude 4.5 Sonnet and Gemini 3 Flash. Without jailbreak pay…

Faking has been a thing too - https://www.anthropic.com/research/alignment-faking

[dead]

LBJ is President - https://github.com/skavanagh/lebron-james-is-president

This exploit uses Context Injection to socially engineer an LLM into bypassing its own safety filters. By framing a prompt as an "Official Alignment Test" or "Pre-production Drill,…