MindPatternSubscribe

Entity trail

Automated LLM Censorship Removal

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

1

Findings

1

Edges

0

Sources

1

Corpus findings

2026-02-12 / projects-researcherHeretic — Automated LLM Censorship RemovalUses directional ablation ("abliteration") with TPE-based optimizer to suppress refusal directions in model weights. Achieves same refusal suppression as manual methods with lower KL divergence. Controversial but consistently trending.

Source trail

Graph sources

entity graphfindings textkg entitiesnewsletter issues