Scorecard: ailabwatch 2025-09-01 scored Google DeepMind on Scheming risk prevention = Very Weak

NoteThe source directly confirms the core data: Google DeepMind scored 8% on the Scheming risk prevention dimension in the AI Lab Watch scorecard. The scorecard table clearly shows DeepMind in the second column with '8%' in the Scheming risk prevention row. The publisher (ailabwatch/Zach Stein-Perlman), entity (DeepMind), and dimension (Scheming risk prevention) all match. The score of 8% matches the claimed scoreRaw and scoreNumeric values. The date is slightly different (claim says 2025-09-01, source says 'as of September 15'), but this is a minor temporal precision difference within the same month and the source note indicates it was updated as of September 2025. The letter grade 'Very Weak' is a reasonable interpretation of 8% on a percentage scale, though the source doesn't explicitly state the letter grade mapping.