Google and the Pentagon have signed a new contract more permissive than those of rival companies like OpenAI. The agreement allows Google's Gemini model to be used for classified settings and allows the company to adjust its safety settings at the government's request. However, critics argue that this agreement relies on contract language that is not legally binding. The deal comes as AI labs develop more powerful models and the technology's use comes under scrutiny amid the Iran war. Outside groups are hoping this will increase efforts to include military AI safeguards in the annual defense policy bill. They are calling for greater transparency around AI company dealings with the Pentagon and modernizing the AI testing and verification process.