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

NoteThe record claims Google DeepMind scored 8% (Very Weak) on Scheming risk prevention as of 2025-09-01. The source is the AI Lab Watch scorecard which explicitly lists 'Scheming risk prevention' as a dimension with scores for multiple companies. The second column in that row shows '8%' which corresponds to DeepMind (the second company listed in the overall scores: Anthropic 28%, DeepMind 20%, OpenAI 18%, etc.). The date is consistent (September 2025, with the source noting it was last updated September 15, 2025). The scoreRaw value of 8% matches the extracted value. The scoreLetter 'Very Weak' is a reasonable qualitative interpretation of 8% on a percentage scale. All key fields are confirmed by the source.