A class action was filed against OpenAI on Tuesday.
Plaintiff: Amargo Couture. Southern District of California. Case No. 3:26-cv-03000.
The allegation: ChatGPT.com has been embedding Facebook Pixel and Google Analytics tracking codes that send your chat topics, user IDs, and email addresses to Meta and Google in real time. Without consent.
OpenAI hasn't responded yet. The case will play out in court for months.
But the fixes take 10 minutes. And they're good practice whether the lawsuit wins, loses, or settles.
Why most people get this wrong:
They wait.
They tell themselves they'll "look into it later." Later never comes.
Meanwhile every chat you typed today, last week, last year, is potentially sitting on OpenAI's servers, in their Memory feature, and (per the lawsuit) in Meta and Google's advertising systems.
You can't undo what you already shared. But you can stop the bleeding tonight.
Here are the 5 steps. Each one takes under 2 minutes.
1/ Kill Training
Path: Settings → Data controls → "Improve the model for everyone"
Action: Toggle OFF.
Why: This is the master toggle. With it on, OpenAI uses your conversations to train future ChatGPT models. The lawsuit alleges your queries are also shared with third parties via embedded trackers regardless of this toggle, but turning it off cuts OpenAI's training use immediately. Start here.
2/ Block Trackers
Path: Install uBlock Origin (Chrome/Firefox) OR switch to the Brave browser.
Action: Visit ChatGPT.com, click the extension shield, watch how many trackers it blocks.
Why: The lawsuit alleges ChatGPT embeds Facebook Pixel and Google Analytics. uBlock Origin blocks both. This is the only step that stops the leak at the browser level. OpenAI account settings can't stop trackers from firing once you land on the page.
3/ Wipe Memory
Path: Settings → Personalization → Memory → "Manage"
Action: Review every stored entry. Delete anything sensitive. Toggle Memory off entirely if you don't trust it.
Why: Memory stores facts about you across chats. Jobs. Addresses. Health info. Relationships. Salaries. The lawsuit names this category of stored personal data as part of what may be exposed. Wipe what you don't want sitting in OpenAI's database.
4/ Delete History
Path: Settings → Data controls → "Manage" → Delete all chats
Action: Export your data first (Data controls → "Export data") if you want a backup. Then delete all.
Why: Past chats sit on OpenAI's servers indefinitely unless you actively delete them. Even after deletion they're retained for 30 days. Delete anything you wouldn't want in a future breach or subpoena.
5/ Temp Chat Mode
Path: ChatGPT composer → click the model selector → toggle "Temporary chat"
Action: Use Temporary Chat for any query involving finances, health, legal questions, or other people's information.
Why: Temporary Chats don't save to history, don't update Memory, and aren't used for training. They're deleted after 30 days. Make this your default for anything sensitive. The incognito mode of ChatGPT.
Setup (10 minutes):
Steps 1, 3, 4 happen inside ChatGPT settings. Do them in one sitting.
Step 2 is a browser install. Takes 60 seconds.
Step 5 is a behavior change. Set Temporary Chat as your default mindset for anything you wouldn't say on a podcast.
The compound effect:
Tonight: Your account is locked down. Future queries no longer feed training. Sensitive chats stay temporary.
Week 1: You stop typing things into ChatGPT you wouldn't want subpoenaed.
Month 1: You realize how much you'd been treating ChatGPT like a therapist, lawyer, or doctor. You start using Claude (which has stronger privacy defaults) for the truly sensitive stuff, and ChatGPT for the public-facing work.
Year 1: When the next AI privacy lawsuit drops (and it will), you're already protected.
That's the shift.
Most people share everything with AI assuming it stays private. You're going to be the one who knew better.
Do this tonight. 10 minutes. That's it.
While everyone else is finding out about this from a news headline next month, you'll already have your account locked down.
Cheers,
Matas Jonaitis
Founder, DailyChatGPT (@dailyprompter)