How to delete your data from AI: OpenAI, Slack, Adobe, Google…

Absolutely everyone has posted something on the Internet in the last 20 years, so it is very likely that artificial intelligence companies have absorbed and used our information to help train the current wave of generative AI.

ChatGPT DOWNLOAD

Large language models, like ChatGPT, and image creators thrive on large amounts of our data. And even if it’s not a chatbot, the data can be used for other machine learning functions.

Technology companies have scoured vast swathes of the web to collect data they claim is necessary to create generative AI, with little regard for content creators, copyright laws, or privacy.

And to add to the problems, companies that have tons of messages from people (like Reddit, Twitter, etc.) are trying to get a piece of the AI cake by selling or licensing that information.

However, as demands and investigations around generative AI and its opaque data practices continue to accumulate, there have been small advances to give people more control over what happens with what they post online.

Some companies now allow individuals and businesses to choose not to have their content used for AI training or sold for training purposes. This is what you can and cannot do.

How far can we go to stop AI training

Before discussing how you can opt out, it is important to set some expectations. Many companies that create AI have already crawled the web, so it is likely that everything you have posted is already in their systems.

In Wired, an expert explains that companies can make it difficult to opt out of having their data used for AI training, and even when it is possible, many people do not have a “clear idea” about the permissions they have agreed to or how the data is being used.

And that’s before taking into account various laws, such as copyright protection laws and strict European privacy laws. Facebook, Google, Twitter, and other companies have included in their privacy policies that they can use your data to train AI.

Although there are several technical ways to delete or “unlearn” data from AI systems, says the expert, very little is known about the existing processes. The options may be buried or require a lot of work. Getting messages to be deleted from AI training data will probably be an arduous battle.

In cases where companies are starting to allow opt-out for scraping or data sharing in the future, they almost always make users accept it by default.

Although less common, some companies that create AI tools and machine learning models do not automatically include customers.

How to delete our data from these services

Adobe

If you store your files in Adobe Creative Cloud, the company can use them to train its machine learning algorithm. “When we analyze your content for the purpose of improving and developing products, we first aggregate your content with other content and then use the aggregated content to train our algorithms and improve our products and services,” says the company’s FAQ. This does not apply to files stored only on your device.

If you use a personal Adobe account, it’s easy to opt out. Open the Adobe privacy page, scroll down to the Content Analytics section, and click on the button to deactivate it. For business or educational accounts, the deactivation process is not available at an individual level, so you will need to contact your administrator.

Google: Gemini

For users of Google’s chatbot, Gemini, conversations can sometimes be selected for review by humans in order to improve the AI model. However, opting out is easy.

Open Gemini in your browser, click on Activity and select the Disable dropdown menu. Here you can simply disable Gemini Apps Activity, or you can choose to opt out as well as delete your conversation data.

While this means in most cases that future chats will not be seen for human review, the selected data is not erased through this process. According to Google’s Privacy Center for Gemini, these chats can remain for up to three years.

OpenAI: ChatGPT and Dall-E

People reveal all kinds of personal information when using a chatbot. OpenAI offers some options about what happens with what you tell ChatGPT, including allowing their future AI models to not be trained with the content.

In its help pages, OpenAI says that web users of ChatGPT without an account should go to Settings and uncheck the option Improve the model for everyone. If you have an account and are logged in through a web browser, select ChatGPT, Settings, Data Controls, and then disable Chat history and training.

If you use the ChatGPT mobile apps, go to Settings, select Data Controls, and disable Chat History and Training. According to OpenAI support pages, changing these settings will not sync across different browsers or devices, so you will need to make the change in all the places where you use ChatGPT.

OpenAI is much more than ChatGPT. For their image generator Dall-E 3, the startup has a form that allows you to submit images to be removed from “future training datasets”.

Slack

All those random Slack messages at work could also be used by the company to train their models.

Although the company does not use customer data to train a large language model for its Slack AI product, Slack may use your interactions to improve the machine learning capabilities of the software.

The only real way to unsubscribe is for your administrator to send an email to Slack at feedback@slack.com. The subject of the message should be “Slack Global model opt-out request” and include the URL of your organization. Slack does not indicate how long the exclusion process takes, but it should send you a confirmation email once completed.

WordPress

Just like Tumblr, WordPress has an option to “prevent sharing with third parties”. To activate it, visit your website’s control panel, click on Settings, General, and then on Privacy, select the Prevent third parties from sharing checkbox.

Your own website

If you host your own website, you can update your robots.txt file to instruct AI robots not to crawl the pages. Most news websites do not allow their articles to be crawled by AI robots.

This exclusion is not only for large media: any website, big or small, can modify its robots file to exclude AI crawlers.