Scorecard: saferai 2025-10-01 scored Google DeepMind on Overall = Very Weak

NoteThe source directly confirms all key fields: (1) publisher is SafeRAI (evident from the scorecard title and methodology); (2) publishedAt is October 2025 (stated as 'Up to date as of October 2025'); (3) entity is Google DeepMind (visible in the comparison table); (4) dimension is Overall (shown in the 'Overall score' row); (5) scoreRaw is 20% (fifth column in the Overall score row displays '20%'); (6) scoreNumeric is 20 (equivalent to 20%); (7) scoreLetter 'Very Weak' is a reasonable qualitative interpretation of a 20% score on a risk management scale where higher is better. All numeric and categorical values match the source data.