AI web browsers have been making waves, with names like Copilot Mode in Edge, Google's Gemini, and OpenAI's Atlas, among others. But here's where it gets controversial: despite the hype, I'm not convinced they're ready for prime time just yet. I've given them a go, and while they have potential, there are some serious flaws that need addressing. So, let's dive into why I'm holding off on making the switch.
The AI Agent Conundrum
One of the key selling points of these browsers is their AI agents, which are essentially your personal assistants within the browser. They can automate tasks, generate media, and translate content, all with a simple click. However, in practice, I've found these agents to be more of a hindrance than a help.
Agents often get confused by ads and CAPTCHAs, stall, or take an eternity to complete simple tasks. Even when they do work, there are still issues. For instance, I've seen agents pick the first recipe result without considering if it's the best or if I already have some ingredients. Or, they might add ingredients that aren't even in the recipe! To avoid these problems, you have to constantly monitor the agent, which kind of defeats the purpose of having an automated assistant.
The problem is, it's nearly impossible to train an agent to handle every possible action on every website. And they're not just competing with each other; they're up against native AI solutions that are becoming standard on many sites and services. Take Google Sheets, for example. AI agents struggle with prompts related to Sheets, but why bother optimizing an agent for this when Sheets' own Gemini integration does the job better?
Your Privacy, Their Problem
Even if the AI features worked flawlessly, there's the issue of privacy. AI web browsers need access to your browsing data and even your passwords to function. Some, like OpenAI's Atlas, use this data to train their models by default. Many browsers lack proper safeguards, like Chrome's Gemini, which can access and respond to content in shared tabs, including password manager tabs.
AI web browsers also introduce unique vulnerabilities, like prompt injection attacks, where the browser fails to distinguish between malicious code and regular content. Even browsers with decent privacy policies, like Opera Neon, still require you to hand over a significant amount of personal data. With data leaks and hacks being common, is it worth putting your valuable information at risk?
Not Worth the Price Tag
Some AI web browsers are free, like Comet, which offers a $5/month journalism subscription. Others, like OpenAI's Atlas, require a premium ChatGPT subscription to access agent functionality. Then there's Opera Neon, which costs a whopping $20/month and isn't part of a larger chatbot service.
When you consider the limitations of these agents and the privacy concerns, it's hard to justify the cost. I can't imagine paying $20/month for what these browsers currently offer. Even if they improve, it's a steep price for yet another subscription.
Should You Make the Switch?
If you already pay for an AI service with a web browser or use a service like Perplexity, which offers most features for free, it's worth trying out. But none of these browsers are worth paying for on their own. Agentic features need more development time, and there's no guarantee they'll ever be perfect.
Realistically, AI agents will always have limitations, and it's unlikely that AI web browsers will become significantly cheaper or more private anytime soon. For me, traditional browsers like Chrome, Edge, or Firefox, which continue to add free AI features, are a better choice. You can try out the AI features or simply stick to the basics - it's your choice.
