The blog post critiques Semgrep's argument suggesting that model weights hinder auditability in AI supply chains. While Semgrep raises important issues about provenance and verification, the author believes their conclusions are flawed, specifically regarding reverse engineering of model weights and the implications for open weights in AI.