Particle.news

Norway’s $2.2 Trillion Wealth Fund Uses AI to Speed ESG Risk Checks

NBIM says large language models now surface potential misconduct within a day of investing, enabling faster decisions on new holdings.

Overview

  • The fund disclosed the expanded use of AI in its annual responsible investment report published Thursday.
  • NBIM’s ESG team began using Anthropic’s Claude in November 2024, then in 2025 rolled out large language models to screen every new entrant to its equity portfolio on day one.
  • The manager receives daily AI-generated risk assessments for prior-day purchases and has, in multiple cases, exited positions before broader market reactions, avoiding potential losses.
  • AI screening has proved most useful for smaller companies in emerging markets by drawing on local-language sources often missed by data vendors and international media.
  • The operational shift comes during a government-led review of the fund’s ethical framework with temporary guidelines that limit exclusion and observation decisions following 2025 divestment controversies.