With Perplexity’s Comet creating buzz, are AI browsers the future or a privacy minefield?
While AI browsers promise instant answers and streamlined experiences, they raise new risks around privacy, misinformation.
- Jul 15, 2025,
- Updated Jul 15, 2025 2:51 PM IST
Artificial intelligence (AI) has already transformed how people interact with the internet. Now, that evolution is accelerating with the launch of Perplexity’s AI-powered browser, Comet, and OpenAI’s plans to unveil its own web browser soon.
While Google Chrome began introducing generative AI features last year and recently rolled out a new “AI mode” experts say the next big leap is moving from search-based browsing to direct, conversational answers. But as these AI browsers promise speed and convenience, questions remain about their impact on user experience, privacy, and the risks of relying on systems trained on potentially biased data.
“The new gateway to the web is no longer search, but AI-driven answers,” says Jaspreet Bindra, founder of AI & Beyond, speaking to Business Today. “Perplexity’s Comet and OpenAI’s planned browser reflect this emerging architecture, where AI becomes the interface to the internet.”
While traditional browsers like Chrome, Safari, and Edge aren’t likely to disappear, Bindra believes they’ll need to evolve rapidly and embed AI capabilities to stay relevant.
“As for who will win the AI browser race, startups like Perplexity have agility and clarity of focus, but giants like OpenAI, Google, and Microsoft have scale, computing power, and an established user base,” he adds.
This shift comes as Google’s global search market share has dipped below 90% for the first time — a signal that users are moving away from traditional searching toward getting direct answers. Tech giants are responding swiftly. In the US, Google Workspace now integrates AI into Chrome, Microsoft has embedded Copilot in Edge, and Apple recently announced “Apple Intelligence” for Safari. Analysts expect these features to roll out globally in the months ahead.
“Google, with its Tensor Processing Units (TPUs), has a significant advantage in terms of raw processing power, which is crucial for AI,” says Puneet Pandey, founder and CIO at Ojas Softech and Astrosage AI.
However, Pandey warns that AI browsers introduce significant new privacy concerns.
“Many AI browsers effectively act as a proxy, sending every webpage you read and every character you type to a third-party LLM for processing,” he says. “Some companies, like Brave, are tackling this head-on by using anonymous reverse-proxy relays and pledging ‘no-logging, no-training’ policies on user data. But most mainstream vendors haven’t offered this level of transparency. The idea that your entire browsing history could become training data for a large corporation represents a fundamentally new privacy risk.”
Beyond privacy, there’s also the risk of misinformation. Bindra cautions that if large language models (LLMs) hallucinate or misinterpret context, false information could proliferate quickly.
Pandey, however, believes hallucinations are becoming less frequent as AI models improve. Still, he sees broader consequences for the digital ecosystem.
“In my opinion, the more immediate threat of misinformation comes from AI-powered image and video generators, which can create highly realistic but entirely fake content,” Pandey says. “But there’s another side effect. We’ve seen publisher click-through rates collapse from 7.3% to just 2.6% where AI summaries appear. Users get a single, potentially flawed answer, and original sources lose traffic and revenue. The only remedy is rigorous citation and easy ways for users to view source material. But the fundamental question of liability for misinformation remains far from settled.”
As AI browsers collect vast troves of user data and reshape how people explore the web, experts agree that new privacy protections and consent frameworks will be essential.
“It’s a powerful shift,” says Bindra, “but one that demands equal measures of responsibility and innovation.”
Artificial intelligence (AI) has already transformed how people interact with the internet. Now, that evolution is accelerating with the launch of Perplexity’s AI-powered browser, Comet, and OpenAI’s plans to unveil its own web browser soon.
While Google Chrome began introducing generative AI features last year and recently rolled out a new “AI mode” experts say the next big leap is moving from search-based browsing to direct, conversational answers. But as these AI browsers promise speed and convenience, questions remain about their impact on user experience, privacy, and the risks of relying on systems trained on potentially biased data.
“The new gateway to the web is no longer search, but AI-driven answers,” says Jaspreet Bindra, founder of AI & Beyond, speaking to Business Today. “Perplexity’s Comet and OpenAI’s planned browser reflect this emerging architecture, where AI becomes the interface to the internet.”
While traditional browsers like Chrome, Safari, and Edge aren’t likely to disappear, Bindra believes they’ll need to evolve rapidly and embed AI capabilities to stay relevant.
“As for who will win the AI browser race, startups like Perplexity have agility and clarity of focus, but giants like OpenAI, Google, and Microsoft have scale, computing power, and an established user base,” he adds.
This shift comes as Google’s global search market share has dipped below 90% for the first time — a signal that users are moving away from traditional searching toward getting direct answers. Tech giants are responding swiftly. In the US, Google Workspace now integrates AI into Chrome, Microsoft has embedded Copilot in Edge, and Apple recently announced “Apple Intelligence” for Safari. Analysts expect these features to roll out globally in the months ahead.
“Google, with its Tensor Processing Units (TPUs), has a significant advantage in terms of raw processing power, which is crucial for AI,” says Puneet Pandey, founder and CIO at Ojas Softech and Astrosage AI.
However, Pandey warns that AI browsers introduce significant new privacy concerns.
“Many AI browsers effectively act as a proxy, sending every webpage you read and every character you type to a third-party LLM for processing,” he says. “Some companies, like Brave, are tackling this head-on by using anonymous reverse-proxy relays and pledging ‘no-logging, no-training’ policies on user data. But most mainstream vendors haven’t offered this level of transparency. The idea that your entire browsing history could become training data for a large corporation represents a fundamentally new privacy risk.”
Beyond privacy, there’s also the risk of misinformation. Bindra cautions that if large language models (LLMs) hallucinate or misinterpret context, false information could proliferate quickly.
Pandey, however, believes hallucinations are becoming less frequent as AI models improve. Still, he sees broader consequences for the digital ecosystem.
“In my opinion, the more immediate threat of misinformation comes from AI-powered image and video generators, which can create highly realistic but entirely fake content,” Pandey says. “But there’s another side effect. We’ve seen publisher click-through rates collapse from 7.3% to just 2.6% where AI summaries appear. Users get a single, potentially flawed answer, and original sources lose traffic and revenue. The only remedy is rigorous citation and easy ways for users to view source material. But the fundamental question of liability for misinformation remains far from settled.”
As AI browsers collect vast troves of user data and reshape how people explore the web, experts agree that new privacy protections and consent frameworks will be essential.
“It’s a powerful shift,” says Bindra, “but one that demands equal measures of responsibility and innovation.”