Florida AG asks court to bar OpenAI from training new models
A 38-page motion also targets ChatGPT for minors, engagement 'dark patterns' and 'human-like' marketing — a regulatory shot at your OpenAI stack.
Copy markdown
What the motion actually demands
Filed Sept 28 in a Florida circuit court, the 38-page emergency motion asks a judge to bar OpenAI from developing models without independent third-party safety oversight, to cut Florida minors off ChatGPT, and to drop features built to extend engagement. If a judge grants even part of it, it's the first court-ordered brake on a frontier lab's roadmap.
A ban on 'human-like' marketing
The motion would block OpenAI from marketing ChatGPT as having 'human characteristics, emotions, consciousness or independent thoughts,' or as 'safe, reliable or accurate' without warnings. Any product you've shipped around a persona or anthropomorphic UX just picked up a legal-risk flag.
The COPPA and dark-patterns claims
The underlying suit, filed June 1 under Florida's Deceptive and Unfair Trade Practices Act, alleges under-13 data collection in violation of COPPA and 'dark patterns' that stretch out conversations. The injunction escalates the case before discovery has even opened.
Why this lands on your stack
The training-halt ask is a long shot, but age-gating and COPPA orders aren't — and a win in Florida becomes the template other state AGs copy, flowing down to any app you build on the OpenAI API. OpenAI says it wants 'pragmatic AI policies that apply to the entire industry — not just one company.'