Detection Is Cheap, Routing Is Learned: Why Refusal-Based Alignment Evaluation Fails

arXiv:2603.18280 · cs.LG, cs.AI, cs.CL · Submitted 2026-03-18 · Read on arXiv

cs.LG, cs.AI, cs.CL

Submitted: 2026-03-18

Updated: 2026-10-20

Comments: Code and data: https://github.com/gregfrank/routing-is-learned

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Ablation — 2× in this paper · explained in 1 episode
• Accuracy — 1× in this paper · explained in 1 episode
• Benchmark — 1× in this paper · explained in 2 episodes
• Cross-Model Transfer — 1× in this paper · explained in 2 episodes
• Generalization — 1× in this paper · explained in 13 episodes
• Steering — 1× in this paper · explained in 2 episodes

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks