No Frontier AI Lab Scores Above C+ on Security Controls

A nonprofit assessment of Anthropic, Google, Meta, OpenAI and xAI found none scored above a C+ on AI control practices. Security practitioners want evidence, not assurances.