Scorecard: ailabwatch 2025-09-01 scored Google DeepMind on Prep for extreme security = Very Weak

NoteThe source directly confirms the core data: Google DeepMind scored 5% on the 'Prep for extreme security' dimension in the AI Lab Watch scorecard. The scorecard is published by ailabwatch (Zach Stein-Perlman's AI Lab Watch). The scoreRaw value of 5% matches exactly. The scoreLetter 'Very Weak' is a reasonable qualitative interpretation of a 5% score, though the source doesn't explicitly use this label. The date discrepancy (claim says 2025-09-01, source says 'as of September 15') is minor and within the same month; the source note states 'Update: as of September 2025, I'm no longer maintaining this website,' suggesting the data shown reflects the final state before that date. All key fields are confirmed or reasonably inferred from the source.