LinkedIn Is Using Your Data to Train AI (but You Can Stop It)
Generative AI models don’t come out of a vacuum. In a sense, these systems are built piecemeal using vast amounts of training data, and they always require more and more information to continue to improve. As the artificial intelligence race heats up, companies are doing everything they can to feed their models more data —and many are using our data to do so, sometimes without asking our explicit permission first.
LinkedIn is the latest apparent culprit of this practice: Everyone’s “favorite” career-focused social media platform appears to be using our data to train its AI models without asking permission or disclosing the practice first. Joseph Cox of 404Media originally reported this story , but you don’t have to be a journalist to investigate it yourself.
This tweet is currently unavailable. It may be downloading or has been deleted.
Simply go to LinkedIn, click on your profile and select Settings & Privacy > Data Privacy . You’ll notice an interesting box here: Data to Improve Generative AI . This option asks, “Can LinkedIn and its affiliates use your personal data and the content you create on LinkedIn to train generative AI models that create the content?” Oh, what is this? Is it enabled by default ? Thanks for the question, LinkedIn.
What does LinkedIn use your data for?
If you click the Learn More link, you’ll see LinkedIn’s explanation of what it does with your data . When enabled, your profile data and the content of your posts can be used to train or fine-tune generative AI models on both LinkedIn and its affiliates. Who are these branches? Well, LinkedIn says that some of their models are provided by Microsoft Azure OpenAI, but as far as I can tell they don’t go into detail.
In this explanation, the company emphasizes that only AI models trained to generate content, such as its AI writing assistant, use your data, not AI models responsible for personalizing LinkedIn for each user, or those used for security . It also says its goal is to “minimize” personal data used in training sets, including by using “privacy-enhancing technologies” to hide or remove personal data from those databases, but does not specify how it does this or to what extent? However, they do offer a form to opt out of having your data used for “non-content generating GAI models.” So what is it, LinkedIn?
Interestingly, Adobe took the opposite approach when users complained about its policies on accessing user work for training AI models: they were adamant that user data was not used for generative AI models, but for other types of AI models. In any case, these companies don’t seem to understand that people would prefer to have their data excluded from all AI training sets, especially if they weren’t asked about it in the first place.
LinkedIn says it keeps your data for as long as you do: If you remove your data from LinkedIn by deleting a post or using the LinkedIn data access tool , the company will delete it on its end and thus stop using it for AI training . The company also clarifies that it does not use user data in the EU, EEA and Switzerland.
In my opinion, this practice is ridiculous. I think it’s unconscionable to allow your users to train AI models on their data without giving them that option first, even before updating the terms of service. I don’t care that lax data privacy laws allow companies like LinkedIn to store everything we post or upload on their platforms: if you want to use someone’s posts to improve your writing bot, ask them first.
I reached out to LinkedIn specifically asking about some of the inconsistencies in its policies and asking how long the process was taking. The company responded to my queries with a link to a page with updates to the user agreement and privacy policy, but the former does not appear to be currently being updated. LinkedIn says the proposed changes will go into effect on November 20, but the current user agreement runs through 2022. Additionally, the support articles I link to in this article were updated seven days ago at the time of writing.
How to Opt Out of LinkedIn’s Artificial Intelligence Training
To continue using LinkedIn without sharing your data to train AI models, return to Settings & Privacy > Data Privacy > Data for Generative AI Improvement . Here you can click the switch to the Off position to opt out. You can also use this form to “object or request a restriction” of the processing of your data for “GAI non-content generating models.”
This is not retroactive: any training already performed cannot be undone, so LinkedIn will not remove training effects from your data from its models.
If you opt out, your data may still be used for AI processing , but only when you interact with AI models: LinkedIn says it may use your input data to process your request and include any data from those inputs in the output AI. , but that’s exactly how AI models work. If LinkedIn can’t access this data after you opt out, the model will be virtually useless.