GPT: definition and legal use | Legaltech Glossary

GPT (Generative Pre-trained Transformer)

GPT (Generative Pre-trained Transformer) is a family of AI models developed by OpenAI, built on the Transformer architecture introduced in 2017. These generative models are pre-trained on vast text corpora and then fine-tuned for specific tasks. The latest version, GPT-4o, is multimodal and powers ChatGPT, among other products.

OpenAI's GPT family brought generative artificial intelligence to the mainstream. The acronym stands for Generative (it generates text), Pre-trained (pre-trained on a vast corpus) and Transformer (a neural architecture built around attention mechanisms). From GPT-1 in 2018 to GPT-4o in 2024, each generation has delivered significant gains in coherence, reasoning and multimodal capabilities.

In the legal sector, GPT powers professional solutions such as Harvey AI, used by the law firm Allen & Overy. Tools like these enable case-law research, drafting of memos and the summarisation of complex documents. When used on its own through ChatGPT, however, GPT is not connected to verified legal databases and carries a risk of hallucinations.

For lawyers and in-house counsel, it is essential to distinguish the consumer use of ChatGPT from professional solutions that embed GPT with safeguards designed for the law: RAG over certified databases, temperature control, and systematic human review of every output.

Related terms