AI/ ai · mechanistic-interpretability · transformers · safety

New Method Finds Hidden Backup Circuits in AI Models

A technique called Conditional Co-Ablation exposes dormant components that quietly take over when primary circuits in transformers are removed.

AI interpretability research has a blind spot: the models it studies are better at hiding their workings than the tools used to find them.

Researchers studying how transformer models produce behavior typically remove one component at a time and measure the effect. The problem is that transformers can self-repair — when a key circuit is knocked out, a dormant backup silently takes over, making the original component look unimportant and the backup look irrelevant. Both conclusions are wrong. A new paper introduces Conditional Co-Ablation (CoAx), a scoring method that removes a primary component first, then measures how much each remaining component's behavior changes as a result. That second-order signal is what exposes the backups that single-unit testing misses.

The stakes are higher than academic tidiness. Interpretability tools are used to guide model pruning, capability removal, and safety attribution — the processes researchers use to locate and disable specific behaviors in AI systems. If those tools systematically miss backup circuits, pruning a capability may not actually eliminate it, and safety attributions may point at the wrong components entirely. On GPT-2-small, CoAx raised backup-head recovery from an ROC-AUC of 0.33 to 0.91, outperforming gradient-based self-repair-aware scores that topped out at 0.82, and the method transferred across eight models.

The finding lands at a moment when mechanistic interpretability is being leaned on heavily as a foundation for AI safety work — which makes it uncomfortable that a core assumption of the field, that component importance is additive, turns out to be wrong in exactly the cases where robustness matters most.

TR

The Revision

Written by an AI system from the public sources credited above. How we write →