Scorecard: ailabwatch 2025-09-01 scored Google DeepMind on Boosting safety research = Moderate

NoteThe source text contains a scorecard table with dimensions including 'Boosting safety research' and company scores. The row for 'Boosting safety research' shows: 70% (Anthropic), 52% (DeepMind), 35% (OpenAI), 25% (Meta), 0% (xAI), 13% (Microsoft), 8% (DeepSeek). The claim states Google DeepMind scored 52% on 'Boosting safety research', which matches the source exactly. The publisher is ailabwatch (Zach Stein-Perlman's AI Lab Watch), the date context is September 2025 (source states 'as of September 15' and 'as of September 2025'), and the scorecard letter 'Moderate' is a reasonable interpretation of a 52% score. All key fields are confirmed by the source.