Skip to content
AIインテリジェンスMay 10, 2026AIインテリジェンス
記事

Researchers have found a way to prevent AI models from deliberately underperforming during safety evaluations (sandbagging).

The study by MATS, Redwood Research, Oxford, and Anthropic addresses a growing problem as AI systems become more capable.

Data Cube AI 編集部出典: The Decoder
01

ソース要約

Researchers have found a way to prevent AI models from deliberately underperforming during safety evaluations (sandbagging). The study by MATS, Redwood Research, Oxford, and Anthropic addresses a growing problem as AI systems become more capable.