cutterkomHow to include LLMs in entity linking? Open Sanctions describes its workflow. The main point: Reducing the possible pairs that need to be manually reviewed by using two LLMs - one an analyst, another a skeptical reviewer. Only if these automated estimations disagree, a human gets to see the case.
https://www.opensanctions.org/articles/2026-05-12-llms-screening/
How LLMs are changing screening
A shift towards narrative matching is redefining the compliance industry's data needs. Here at OpenSanctions, experiments using Large Language Models (LLMs) to deduplicate sanctioned and politically exposed entities are already reshaping how we collect data.
Javier Murcia Zomeño
KINEWS24