AI labs agreed to slow development and allow third-party safety reviews. They will grant reviewers employee-level access and let them publish findings. The White House still downplays risks to avoid falling behind China. Congress is urged to create an AI self-regulatory organization. Financial, healthcare and life sciences firms should demand security assurances from AI vendors. Open-weight models pose a diffusion challenge that needs industry coordination. An alliance led by Nvidia should identify and fix vulnerabilities in open-weight models. Trust in AI requires enforceable standards and independent oversight.