Refusal Beyond a Single Direction: A Preliminary Comparison of Diff-in-Means and

<!-- SC_OFF --><div class="md"><p>I set out to test whether AAVE-coded (African American English Vernacular) prompts cause MoE language models to route, deliberate, and respond

arXiv:2605.21706v1 Announce Type: new Abstract: Safety-aligned language models are trained to refuse harmful requests, yet refusal behavior can be suppressed by steering their

arXiv:2608.28233v1 Announce Type: new Abstract: Steering with Sparse Autoencoders (SAEs) offers a lightweight inference-time path for adapting the behavior of large language

arXiv:2608.08032v1 Announce Type: new Abstract: Safety alignment in multilingual models is uneven: a model that reliably refuses a harmful request in English will often comply

arXiv:2603.23406v1 Announce Type: new Abstract: While large language models simulate social behaviors, their capacity for stable stance formation and identity negotiation during

arXiv:2604.06233v1 Announce Type: new Abstract: Safety-trained language models routinely refuse requests for help circumventing rules. But not all rules deserve compliance. When

<a href="https://news.ycombinator.com/item?id=47986136">Comments</a>

arXiv:2608.11392v2 Announce Type: cross Abstract: Long-running agents periodically compact their context, replacing the transcript with a model-generated summary. Recent work