Frontier AI access and research governance moved again this week as Anthropic restored global access to its Fable 5 model after a two-week government standoff that had restricted use of the company’s most powerful models. The administration and Anthropic reached a temporary truce through export-control adjustments tied to alignment considerations, and access was restored for different user groups. At the same time, researchers and observers warned that new AI agents may disrupt how grants are evaluated and awarded, arguing that technology development is outpacing assessment and reform. The concern centers on whether automated or agent-driven evaluation processes could alter how peer review and compliance checks are conducted, potentially changing decision integrity. Together, the stories point to a higher-education-operational reality: universities and research labs must manage compliance with AI access restrictions while rethinking institutional policies for AI-assisted evaluation and grant decision processes.