FTC Warns Against Quiet Privacy Policy Changes as Companies Seek AI Training Data

FTC Warns Against Quiet Privacy Policy Changes as Companies Seek AI Training Data

As companies seek valuable AI data to train new models, they are increasingly turning to private datasets protected by privacy laws, making small changes to privacy statements to allow this use. Earlier this year, the Federal Trade Commission (FTC) warned that companies might quietly amend their privacy statements to use customer data for AI training, which would be illegal and could result in legal consequences. The FTC emphasized that adopting more permissive data practices without clear user consent is deceptive and unfair.

Despite this warning, companies like Google, Adobe, and Meta have altered their privacy policies to include terms like "artificial intelligence," "machine learning," and "generative AI." Google, for instance, updated its privacy policy to state that publicly available information is used to train its AI models, including services like Google Translate, Bard (now Gemini), and Cloud AI. Google claimed this change was merely a clarification and not an expansion of data usage.

Adobe faced significant backlash after notifying users that it could use content created with its Creative Suite to train AI models. Users were unhappy about having to agree to new terms to access their projects, leading to canceled subscriptions and forcing Adobe to clarify the update. Similarly, Meta informed European users that it would use public posts to train AI, but paused these plans after complaints from the European Center for Digital Rights.

In the US, where consumer protections are weaker, it is easier for Meta to gather data from its users. The FTC's future actions remain uncertain as more companies continue to modify their privacy policies to incorporate AI data training.