{"id":44280,"date":"2024-07-30T14:28:38","date_gmt":"2024-07-30T18:28:38","guid":{"rendered":"https:\/\/mjtsai.com\/blog\/?p=44280"},"modified":"2024-07-30T14:28:38","modified_gmt":"2024-07-30T18:28:38","slug":"social-media-ai-training","status":"publish","type":"post","link":"https:\/\/mjtsai.com\/blog\/2024\/07\/30\/social-media-ai-training\/","title":{"rendered":"Social Media AI Training"},"content":{"rendered":"

C. Scott Brown<\/a>:<\/p>\n

\n

Meta is now scraping Facebook posts to train its AI model. While this isn’t surprising on its own, what is surprising is just how difficult Meta is making it for users to opt out of this process.<\/p>\n<\/blockquote>\n\n

Via X Daily News<\/a>:<\/p>\n

\n

Instagram is training AI on your data but makes it nearly impossible to opt out.<\/p>\n<\/blockquote>\n\n

Juli Clover<\/a>:<\/p>\n

\n

Social network X (formerly Twitter) recently activated a setting that gives it permission to train Grok AI on user tweets. All X users are opted in by default, with X failing to notify customers about the change.<\/p>\n<\/blockquote>\n\n

Kimmy Bestie<\/a>:<\/p>\n

You can disable this using the web but it’s hidden. You can’t disable using the mobile app.<\/p><\/blockquote>\n\n

Dare Obasanjo<\/a>:<\/p>\n

I found the outrage about X’s default setting to use user content to train xAI’s Grok to be confusing, at least for public tweets.<\/p>

If you’ve ever posted anything on the public internet, it has been used to train AI. If you do so today, it will as well.<\/p><\/blockquote>\n\n

This is my thinking as well. It’s the same with Reddit and YouTube where the users specifically contributed their content to the Web. This is quite different from the situations with Adobe, Slack, Grammarly, and Zoom where the expectation was that the data was private.<\/p>\n\n

Previously:<\/p>\n