Your Writing Is Probably Already in an AI Training Dataset. Here'\''s How to Opt Out.

#ai #privacy #python #opensource

Your writing is already in a training dataset

Every message you have sent on a platform that later added AI features. Every document you stored in a cloud service that updated its terms of service. Every photo you posted before the AI training clause appeared in the privacy policy. Every forum post, every review, every code snippet on a platform that has a model to train.

The data collection happened quietly, usually through a terms-of-service update you did not read, often retroactively applied to content you created years earlier.

GDPR Article 21 gives you the right to object to processing for purposes including AI training. CCPA gives California residents the right to opt out of the sale or sharing of personal information — which includes training data pipelines. These rights exist. Most companies will honor formal requests because the legal exposure for ignoring them is significant.

The problem is making the requests. Fifteen AI companies, each with different processes.

I built ai-opt-out to automate the tedious part.

The companies covered

ai-opt-out list

Company	Method
OpenAI	Email
Google / DeepMind	Form
Meta AI	Form
Microsoft Copilot	Form
Apple	Form
Adobe Firefly	Email
Amazon / Alexa AI	Form
X / Grok	Form
Anthropic (Claude)	Email
Cohere	Email
Stability AI	Email
Mistral AI	Email
Hugging Face	Email
Spotify AI Features	Form
Netflix AI	Form

Two methods: email and form. Email companies get a legally-formatted draft copied to your clipboard. Form companies get the opt-out URL printed with instructions copied to your clipboard. Either way, you do the final action — ai-opt-out does the preparation and tracking.

Sending opt-out requests

# Send to one company
ai-opt-out send openai --email me@example.com

# Use GDPR Article 21 explicitly
ai-opt-out send google --email me@example.com --gdpr

# Send to all companies at once (with confirmation)
ai-opt-out send --all --email me@example.com --yes

The GDPR Article 21 email template:

Subject: Objection to Processing for AI Training — GDPR Article 21

Dear OpenAI Privacy Team,

I am writing to exercise my right to object under Article 21 of the General
Data Protection Regulation (GDPR).

I object to the processing of my personal data — including any content I have
created, submitted, or otherwise provided — for the purposes of training,
fine-tuning, or otherwise developing machine learning or artificial intelligence
models.

I request that you:
1. Cease using my personal data for AI training immediately
2. Delete any training data derived from my personal information
3. Confirm compliance within 30 days as required by GDPR Article 12(3)

My email address is: me@example.com

[Your name]

Tracking and follow-up

Most companies will respond. Some will not. ai-opt-out tracks every request you send:

ai-opt-out status
ai-opt-out status --email me@example.com

Requests that have gone unanswered beyond the statutory deadline are flagged. When something is overdue, you have documented grounds to escalate — to the company, to a data protection authority, or to a privacy advocacy organization.

A note on realistic expectations

Sending formal opt-out requests does not guarantee your data is removed from existing models. Models that have already been trained on your data cannot have that training undone. What you are doing is:

Creating a legal record that you objected
Preventing future use of your data for training
Exercising rights that are costly to ignore at scale

If enough people send formal requests, the calculus changes for companies. Individual requests are easy to ignore. Systematic, documented requests at scale create legal and regulatory exposure.

Install

pip install ai-opt-out

# See all companies
ai-opt-out list

# Send to one
ai-opt-out send openai --email me@example.com

# Send to all
ai-opt-out send --all --email me@example.com --yes

# Check status
ai-opt-out status