As AI tools become increasingly ubiquitous, it is important to question how they are or are not protecting user data. Technology tools are moving away from the opt-in standard for data privacy to a training by default method, where user information is recorded and applied to fuel the next generation of large language models. Privacy is rarely the default setting in the current AI landscape; instead, the burden is placed on the user to protect themselves.
OpenAI’s Agentic Web Browser
OpenAI’s new web browser takes contextual awareness to a new level. The agentic AI tool can navigate the web and actually take action on your behalf, completing tasks like handling your day-to-day scheduling or purchasing dinner ingredients. But to do so effectively, it watches your every move online. While the convenience of a browser with ChatGPT “baked in” may be appealing, this agency without oversight creates a massive security risk that many users are overlooking.
OpenAI's new web browser has ChatGPT baked in. That's raising some privacy questions (NPR)
Anthropic’s Major Policy Change
Anthropic, the self-described ethical AI firm, updated its policies in August 2025 to require users to manually opt out of data training. Claude is now using previously private conversations to sharpen future models by default. For those who don't catch this change, the standard 30-day data deletion window vanishes, replaced by a five-year retention period that keeps your data in Anthropic’s hands long after you've forgotten the conversation.
Anthropic users face a new choice – opt out or share your chats for AI training (TechCrunch)
Automatic Opt-in Is the AI Default
The Stanford Institute for Human-Centered AI reported that nearly all leading AI developers use conversations to train their models by default. This places the entire burden of data protection on the user to find opt-out settings buried within menus. Researchers warn that these systems are recording sensitive information, often retaining it indefinitely without an obvious path for user deletion.
Study exposes privacy risks of AI chatbot conversations (Stanford Report)