AI Browsers: 5 Crucial Things You MUST Know Before Using One (2024)

Imagine a browser that anticipates your needs, clarifying articles, explaining images, and even handling tasks like online shopping. That’s the promise of the new wave of AI browsers, spearheaded by OpenAI’s ChatGPT Atlas and Perplexity’s Comet. But before diving in, it’s vital to understand the implications. Are these AI helpers worth the potential privacy trade-offs?

## AI Browsers: Convenience vs. Privacy

AI browsers differ significantly from traditional browsers. A persistent chatbot button lets you ask questions about the content you’re viewing, offering on-demand clarification. Agent mode takes it a step further, delegating tasks like editing documents or completing online purchases. However, this convenience comes at a cost. AI browsers analyze the content of the websites you visit, including potentially sensitive information like order histories or messages. This data is then sent to the AI provider’s servers for processing, raising serious privacy concerns.

## Data Control & Model Training: What You Need to Know

With AI browsers, you might have less control over the data shared compared to using a chatbot directly. The browser automatically attaches the website you’re viewing for context. While OpenAI claims their AI is selective about the information it retrieves (image, text, etc.), users lack transparency. To mitigate risks in Atlas, remove pages from the chat or block specific websites. Turning off model training is crucial. Atlas has two settings: “Improve the model for everyone” (ON by default) and “Include web browsing” (OFF by default). Disable the former to prevent your data from being used for AI training.

## The Inherent Risk: Data Security

Even if you opt-out of training, remember that your data is still being sent to the AI provider. As privacy experts warn, once your sensitive data resides on another company’s servers, you relinquish significant control. Misuse by hackers or governments becomes a real possibility. Furthermore, be vigilant about “prompt injection” attacks, where malicious actors can inject hidden instructions into websites that hijack the AI agent to steal sensitive information. Exercise caution when activating agent mode on unfamiliar sites. While AI browsers offer intriguing possibilities, a thorough understanding of their privacy implications is paramount. Consider these points carefully before entrusting your browsing to AI.

AI Browsers: 5 Crucial Things You MUST Know Before Using One (2024)

Related posts: