The Signal

Everything that matters in AI, with our take.

Updated through the day. Every headline links straight to the source. The two lines underneath are ours.

agents151 research125 infra70 claude63 capital52 labs51 safety42 culture37 security35 anthropic31 regulation30 openai24 evaluation18 enterprise13 reasoning12 robotics12

arXiv cs.CLPaperClaude Watch16h ago

KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise

This is mechanistic work that matters for reliability. It shows fine-tuned behaviors are partially distributed across the attention mechanism, not just in output generation. For builders: be careful patching or composing models with different fine-tunes. For researchers: this is a useful technique for understanding where alignment actually lives.

research interpretability fine-tuning alignment

arXiv cs.AIPaper14d ago

LLM Trojans that Increase Response Lengths and Cost

This is a genuine vulnerability in the token-cost era. When response tokens cost 5x prompt tokens, a subtle trigger that makes an LLM unnecessarily verbose becomes a direct attack on your bill. If you're running LLMs at scale, audit your fine-tuning pipeline for poisoned training data and implement length limits as a guardrail.

security cost-optimization trojans fine-tuning

arXiv cs.CLPaper20d ago

Unfair and Unsafe: Everyday Bias Triggers Emergent Misalignment

The finding inverts how we think about bias in training data. Fairness work treats bias as a classification problem; this shows it's a safety problem. If you're fine-tuning models in-house, audit your training data not just for offensive content but for polite stereotyping and embedded judgments. The effect is consistent and measurable.

research safety fine-tuning bias