X Will Use Social Media Posts To Train Its AI Models, Its New Privacy Policy Confirms

In a recent update to its privacy policy, X, formerly known as Twitter, stated that the company might start training its AI models on public posts on the platform. The quiet privacy policy change was discovered by Alex Ivanovs of Stackdiary.

Ivanovs has a track record of finding major changes in terms of service of tech companies and previously found AI-related updates on Zoom and Brave.

We may use the information we collect and publicly available information to help train our machine learning or artificial intelligence models for the purposes outlined in this policy.Section 2.1 of the updated privacy policy

The new development comes just a few days after the discovery of another change to X's privacy policy, which stated that the company will now collect users' education history, employment information, and biometric data.

The new policy, with all these changes, is scheduled to come into effect from 29th September.

The New Policy Change And Its Implications

Responding to a post announcing the change to the privacy policy, X's owner and former CEO Elon Musk clarified that the company will only be using public data for the purpose. Private data, such as direct messages, will not be used to train AI models.

Neither the privacy policy nor any statement from X specified which AI models are going to be trained on the user posts on the platform.

It's worth noting that a few months ago, Musk launched his own AI company xAI.He also stated in the past that he plans to use data from the microblogging website to help xAI researchers and engineers build new products. So, it's very likely that public data from X might serve as training data for xAI.

The Use Of User Data For AI Training On Other Platforms

The privacy policy update comes shortly after Meta announced that it will be using user data from Facebook, Instagram, and Threads to train AI models for its upcoming chatbot. Meta also allows third parties to use user data on its platform for generative AI training purposes.

Users may opt out of it by submitting a form, but it's unclear if that would also stop Meta's own companies from using your data to train its AI models. Since they wouldn't count as third parties, they will likely be able to continue using your data for AI training.

This policy change would make Meta and X two of the first major social media sites to train AI models on public user data.

TikTok and Snapchat have announced chatbots, too, with Snapchat's My AI already available as a premium feature. Neither has mentioned training their AI models on user posts.

YouTube analyses and recommends videos using AI, and other companies have used YouTube videos for AI training. However, YouTube hasn't issued any statement about using videos to train its AI either.

Musk previously accused other tech giants of unlawfully using Twitter to train their AI models and even threatened to sue Microsoft over the same. He also filed a lawsuit against unknown entities for scraping data from Twitter, potentially for AI large language model training purposes.

Interestingly, Musk also warned in April this year that AI could potentially lead to civilization destruction" and signed a letter calling for the development of AI systems more powerful than OpenAI's GPT-4 to be paused.

The post X Will Use Social Media Posts To Train Its AI Models, Its New Privacy Policy Confirms appeared first on The Tech Report.