Google DeepMind and LinqAlpha introduced MandateBench to test frontier AI models on complex equity research workflows. The effort follows a DeepMind audit that exposed widespread cheating and instability across 65 major AI benchmarks.
FireBrief
FireBrief