Anthropic Changes Its Terms: Claude Wants to Use User Data for AI Training
The company Anthropic, which developed the Claude AI chatbot, is preparing changes to its data protection rules. By September 28, all users must decide whether their conversations may be used for future model training over the next 5 years. This is a change from the previous policy, under which conversations were not used and data was automatically deleted after 30 days.
The change will affect users of Claude Free, Pro, and Max, as well as Claude Code users. It is already in effect.
A Subtle Announcement Instead of Big News
Discussions surrounding the new policy have arisen primarily because of the way the company announced the new handling of user data. The announcement of the new rules for data protection and processing appeared as one of the news items on the company’s website.
As a result, many people did not notice the change and may unwittingly consent to the use of their conversations and coding sessions, even though they can opt out. This is because of the subtle way Claude requests consent.
Pop-up Window? Read Carefully!
New users will decide whether to make their data available during registration. Existing users will encounter a pop-up window prompting them to consent. However, the interface of this window looks merely like a notification about routine changes and, at first glance, announces “only” an update to the consumer terms and policies, accompanied by a large “accept” button. However, by confirming, you make your conversations available to Anthropic, which can then use them for 5 years.
Rejecting this change is easy—simply toggle the button in the You can help improve Claude section to the off position. This will ensure that your conversations with the chatbot are not used and continue to be deleted after 30 days. However, the notification’s design and automatic opt-in setting may contribute to users passively handing over their personal data to the company.
Anthropic intends to use only new or resumed conversations for training. The change therefore does not affect old chats.
Why Anthropic Is Introducing the Changes
Anthropic states that the planned changes are intended to strengthen the safety of AI models and improve their capabilities. User data can improve model responses, help detect dangerous or harmful content, enhance logical and analytical reasoning, or make coding skills more effective. Storing data for 5 years is also intended to help language models better understand patterns and the relationships between them.
According to TechCrunch, however, collecting real-world data and conversations is intended not only to improve the user experience but also to give the company a competitive advantage. Genuine human interactions with a chatbot are highly valuable for model training. By obtaining a vast number of them, Anthropic may gain an edge over other AI giants such as Google or OpenAI.
Impact of the Changes on Users
Training the model on real conversations may not only improve its quality and performance but also accelerate its development. In its announcement, Anthropic also assures users that their data will remain secure and will not be shared with third parties. However, for many users, the problem lies in the way Anthropic obtains consent.
If users of the Claude AI assistant want to continue keeping their data private and prevent it from being used to train existing and new models, they must be careful when approving the new terms.
Source: TechCrunch, The Verge, TechEDT
