Max Nadeau Says Method Is Not Mechinterp

Max Nadeau, identified as a Program Officer at Coefficient Giving who funds technical AI safety research on interpretability, robustness, and transparency, made the statement in a post tagged AI Safety. He wrote that the approach in question is not mechinterp. Instead, he described it as just a method that makes use of model internals. The post was addressed to @tautologer, @GuiveAssadi, @tszzl, and @zetalyrae. The evidence shows this as a retweet of his own comment in the visible thread.