Naval: Open-Weight Model Backdoors Would Be Exposed

Naval and Matthew Berman argued that if open-weight models truly contain hidden backdoors or biases, capable closed-source labs would likely be able to detect and publicize them. The debate centered on model trust and the different audit incentives of open versus closed ecosystems.

2026-07-27 ~ 2026-07-27 · 2 related posts

• Naval: hidden backdoors in open-weight models would be found by closed labs — naval · 2026-07-27
• Open-weight models can also expose hidden backdoors, says Matthew Berman in reply to Naval — MatthewBerman · 2026-07-27