Natural language processing
NLPAlso called computational linguistics, text analytics
Natural language processing is a field of artificial intelligence that lets software read, interpret and generate human language. In compliance, it is used to screen adverse media, monitor employee communications, extract obligations from regulation and review contracts. It works on text and transcribed speech, so it falls under data protection, recordkeeping and AI rules.
How it works
NLP tools break text into words and phrases, identify names, places and organisations, classify topics and measure sentiment. A bank might use it to scan news for adverse media on clients, flag possible market abuse in chat messages and emails, or map new rules to internal policies. Large language models now perform many of these tasks, alongside older statistical and rules-based methods.
The technology depends on records that firms must keep. MiFID II requires investment firms to record communications relating to client orders. US broker-dealers must preserve business communications under SEC Rule 17a-4. NLP helps surveillance teams review these volumes. Firms must validate models, handle personal data lawfully under GDPR and document how flagged messages are reviewed.
The rules
Requires investment firms to record telephone conversations and electronic communications relating to client orders and transactions.
EU investment firmsRequires broker-dealers to preserve business records, including communications, in a compliant electronic recordkeeping system. The 2022 amendments applied from 3 May 2023.
SEC-registered broker-dealersGives individuals the right not to be subject to solely automated decisions with legal or similarly significant effects, subject to safeguards including human intervention.
Controllers processing personal data of people in the EUBans certain AI practices, sets risk management, data, documentation and human oversight duties for high-risk AI, and transparency duties for chatbots and generated content.
Providers and deployers of AI systems placed on the EU market or used in the EUCommon questions
How is NLP used in compliance?
Common uses include adverse media screening, communications surveillance, regulatory change mapping, contract review and extraction of data from documents. It helps teams search and classify large volumes of text that staff could not read manually.
What is the difference between NLP and a large language model?
NLP is the broad field of computing with human language. A large language model is one type of NLP technology, trained on very large text datasets, that can generate as well as analyse language.
Does NLP-based surveillance raise data protection issues?
Yes. Monitoring employee communications involves personal data, so firms need a lawful basis, transparency to staff, data minimisation and retention limits under GDPR or UK GDPR, alongside their recordkeeping duties.
