X's New AI Training Terms: How to Protect Your Creative Work

What X Changed in Its Terms and Privacy Policy

On October 16, X published revisions to both its Terms of Service and its Privacy Policy, establishing November 15 as the official effective date. The core change directly affects anyone who shares original content on the platform: X explicitly updated its language to allow outside companies to train artificial intelligence models on user posts.

In the updated Privacy Policy, under the section titled "Third-party collaborators," X explains that it may share or disclose your information with third parties. Unless you manually opt out, those external recipients can use your data for their own independent purposes, specifically including the training of machine learning and generative artificial intelligence models.

At the same time, the Terms of Service document was revised to clarify that by posting content on the platform, you grant X a worldwide, non-exclusive, royalty-free license with the right to sublicense. That license includes the right for X to analyze your text and other information to train its own machine learning and AI models, such as Grok, as well as distribute that data to third parties.

The platform also introduced a strict financial enforcement clause in its legal terms aimed at automated scraping, which means using automated code to download thousands of web pages without permission. Under a new section titled "liquidated damages" (a legal term for a predetermined penalty agreed to in a contract), any entity that accesses or views more than 1,000,000 posts in a 24-hour period in violation of the terms will be liable for $15,000 per million posts.

For independent creators, artists, writers, and educators, these updates change the basic deal of posting on the platform. When you post original photography, design samples, thread breakdowns, or commentary, that work can now be handed to third-party model developers by default. Taking immediate action before the November 15 deadline ensures your creative catalog is not used to train outside generative tools without your clear permission.

How to Opt Out of AI Training Step by Step

X provides an opt-out mechanism within its account settings, but it requires you to find the toggle and turn it off manually. As of October 26, 2026, the primary opt-out controls are accessible on both the desktop browser version and updated mobile apps.

Here is how to disable data sharing for AI training on your account:

1. Open Your Settings Menu

Log in to your account on a web browser or through the mobile app. On desktop, click the "More" icon (the three dots inside a circle) in the left-hand navigation sidebar. On mobile, tap your profile picture in the top left corner. From the menu that expands, select "Settings and privacy".

2. Go to Privacy and Safety

In the settings menu, select the tab labeled "Privacy and safety". This section controls what information you share publicly, how your media is displayed, and how your data interacts with platform partners.

3. Locate the Grok and Third-Party Collaborator Settings

Scroll down until you locate the section labeled "Data sharing and personalization," or click directly on "Grok & third-party collaborators" (the label may appear simply as "Grok" depending on your app version).

4. Uncheck the Data Training Permissions

Look for the checkbox or toggle switch that permits X to use your data. The wording states: "Allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning". Uncheck this box or toggle the setting to the off position.

5. Check Additional Partner Sharing Toggles

While in the "Data sharing and personalization" sub-menu, review the "Data sharing with business partners" tab. If you prefer that your personal browsing habits, location history, and device signals are not shared with commercial ad networks or external analytics partners, toggle this setting off as well.

6. Repeat for Every Account You Run

If you manage multiple handles (such as a personal creator account, a podcast handle, and a brand storefront), this setting does not sync across different profiles. You must switch profiles and manually repeat these steps for every account you control before November 15.

What the Opt-Out Does (and What It Does Not Do)

Turning off these toggles is an essential first step, but it helps to be realistic about the limits of platform privacy settings.

When you turn off the setting:

• X is instructed not to pass your new posts, direct interactions, and Grok prompts into future training datasets.
• Your profile is flagged as opted out for third-party commercial data licensing agreements governed by X's privacy terms.

What the setting cannot do:

• It does not automatically delete training checkpoints that were already compiled and finalized in earlier model versions.
• It does not prevent unauthorized rogue scrapers from viewing public content if they bypass platform safeguards, although the new $15,000 liquidated damages clause gives X stronger legal leverage to sue those scrapers in court.
• It does not remove your public posts from web search engines, which maintain their own independent indexing bots.

Because of these boundaries, updating your settings is only half of the solution. You also need a sensible posting workflow that protects your highest-value creative assets.

Five Practical Ways to Protect Your Creative Work

Adjusting your publishing habits makes it much harder for third-party models to copy your style, voice, or proprietary frameworks. Here are five practical habits you can adopt right away:

1. Shift from Complete Guides to Summary Hooks

If you write detailed, research-heavy threads, avoid posting the entire raw essay natively on the platform. Generative text models ingest cleanly formatted, uncompressed text threads with ease.

Instead, publish a concise summary hook with three or four key takeaways, then link directly to the full piece hosted on your own website, newsletter, or publication platform. This practice accomplishes two things: it shields your long-form prose from bulk platform harvesting, and it drives readers back to an audience channel that you own.

2. Apply Glaze or Visible Signatures to Original Art

Visual artists, comic creators, and illustrators face significant challenges when visual generative tools harvest high-resolution portfolio images. A good habit is to avoid uploading uncompressed full-resolution files directly to public social feeds.

Consider processing your portfolio artwork through protective cloaking tools like Glaze before posting. Glaze adds tiny, subtle mathematical shifts to the pixels that human eyes barely notice, but which disrupt how artificial intelligence systems identify the underlying art style. In addition, place a visible signature or studio watermark across a detailed portion of the composition rather than in an easily cropped bottom corner.

3. Host Digital Templates and Code in Private Repositories

If you create code snippets, spreadsheet formulas, design files, or downloadable templates, do not paste the raw working code or raw formulas directly into public posts. Keep the files inside an external repository or a password-protected delivery link. Share a screenshot showing the output or a short demonstration video of the tool working, and require users to visit your site to access the actual source files.

4. Audit and Archive High-Value Historical Posts

Take an afternoon before November 15 to review your pinned posts, popular threads, and older media. If you previously shared proprietary frameworks, book chapters in progress, or unreleased song stems as public posts, consider archiving or deleting those specific tweets. You can download an entire archive of your historical account data first via "Settings > Your account > Download an archive of your data" so you never lose your own record of what you published.

5. Establish Clear Licensing Statements in Your Profile

While a social media bio does not replace a formal legal contract, stating your copyright terms establishes explicit intent. Adding a brief notice such as "All original writing, art, and video copyrighted; no AI training rights granted without written license" leaves no ambiguity about your commercial expectations. It helps clarify your position if your intellectual property is discovered inside unauthorized corporate data dumps.

Content Protection Checklist: What to Post and Where

Use this quick guide to decide where your content belongs and how to distribute it safely:

Content Type	Training Risk Level	Best Hosting Location	Safety Rule Before Posting
Short Commentary & Hot Takes	Low	Native on X	Post freely; opt-out toggle handles standard collection.
Long-Form Essays & Research	High	Owned blog or newsletter	Post the opening hook on X; direct readers to the external link.
Original Digital Illustrations	High	Personal portfolio site	Downsample resolution; apply pixel cloaking or visible marks.
Behind-the-Scenes Video Clips	Low to Medium	Native video on X	Keep raw source files locally; post edited social cuts.
Code Snippets & File Templates	High	External repositories	Share a demo screen recording; keep raw text off the feed.
Client Case Studies & Data	High	Password-gated client portal	Remove identifying client metrics before sharing public summaries.

Common Mistakes to Avoid

When platform terms change, creators often react with rushed decisions that either harm their growth or fail to protect their rights. Keep these common missteps in mind:

Believing an Opt-Out Deletes What Was Already Harvested

Flipping the opt-out toggle tells the platform to stop including your account in future data-sharing agreements. It does not force an outside company to delete a completed model that was trained months ago. If you want your past data off the web completely, you must manually delete those specific older posts.

Making Your Account Fully Private When You Rely on Reach

Setting an account to private does block automated scrapers, but it also hides your profile from new followers, stops non-followers from seeing your replies, and disables retweets. For most working creators, going private destroys discoverability. A better balance is keeping your account public, turning off third-party sharing in settings, and withholding your uncompressed core files.

Forgetting Team and Secondary Handles

Many creators operate a personal account alongside an e-commerce brand account, a community account, or a support channel. Turning off the AI training toggle on your personal handle leaves your other profiles active and opted in. Set a recurring calendar reminder to log in to every profile you own and confirm the settings match.

Relying on External Bulk Scrapers for Analytics

Because X added a $15,000 liquidated damages penalty per million posts accessed in violation of the rules, third-party analytics tools and unofficial growth scrapers that do not use official application programming interfaces (APIs) face heavy legal risks. Avoid running unauthorized third-party desktop tools that claim to scrape competitor metrics or user follower lists, as these could jeopardize your account status under the revised terms.

Protecting your creative business requires staying informed whenever platforms revise their legal terms. By opting out before the November 15 deadline and keeping your highest-value assets hosted on channels you control, you can continue using social platforms for reach without giving away your creative work for free.

Sources

Checked on October 26, 2026.

• ca.gov
• engadget.com
• aisecret.us
• ainews.com
• pcmag.com
• abplive.com