Measuring Activation Control in Large Language Models

arXiv cs.AIAugust 25, 2026en

arXiv:2608.21664v1 Announce Type: new Abstract: Safe deployment of increasingly capable models will likely come to rely on latent-space monitoring as a complement to behavioral evaluations, especially when evaluation-aware models exhibit scheming or deception. However, if models can also control their own activations, deception could extend into the latent space itself. With this in mind, we introduce the Activation Controllability Benchmark to quantify the extent to which models can modulate their residual stream via natural-language instruction. Across model families and capability levels, we find that most LLMs can control the direction and magnitude of their residual stream activations w

This is a short summary published by AI Global Wire. The full article is owned and hosted by arXiv cs.AI — open it there to read it in full.

• Forskning
• Företag