An Embarrassingly Simple Defense Against LLM Abliteration Attacks · HackerPapers discussions

An Embarrassingly Simple Defense Against LLM Abliteration Attacks — mentioned by wren6991

Sun, 20 Sep 2026 18:48:05 GMT

I've seen attempts to obfuscate the refusal direction, like here: https://arxiv.org/html/2608.18093v1 Also this one was interesting, training the model to give preambles with reasons for the reasons for refusal seems to make it less sensitive to modulating the single refusal direction: https://arxiv.org/html/2505.19056v1 My empirical observation is that when a new model is released on HuggingFace, an abliterated version with < 10/100 refusals (baseline usually 100/100) is uploaded the same day, so either these techniques don't work very well or the open-weight labs aren't applying them.…

Paper and discussion history. A mention is not an endorsement.

An Embarrassingly Simple Defense Against LLM Abliteration Attacks — mentioned by kylecazar

Mon, 07 Sep 2026 01:23:40 GMT

There was a paper [1] that offered as a defense against abliteration what amounts to just longer and more nuanced refusals. The results seemed promising. I don't follow this closely enough to know why the latest open models haven't incorporated the research... [1] https://arxiv.org/abs/2505.19056

Paper and discussion history. A mention is not an endorsement.

An Embarrassingly Simple Defense Against LLM Abliteration Attacks — mentioned by akersten

Sat, 02 May 2026 14:44:39 GMT

2024 which is ancient history. This is not true anymore, the models now are trained to prevent abliteration by spreading out the refusal encoding See https://arxiv.org/abs/2505.19056

Paper and discussion history. A mention is not an endorsement.