Security researchers at Guardio, a cybersecurity firm, have uncovered new vulnerabilities in the latest generation of AI-powered web browsers, raising concerns about how these tools handle online threats.
In their recent security report, they analyze how emerging browser technologies, designed to automate user tasks, are susceptible to sophisticated scams potentially putting users at risk.
Artificial intelligence is transforming our online experience. With the emergence of AI-powered browsers that can autonomously shop, manage emails, and complete tasks for users, convenience has reached a whole new level. But as these digital assistants take on more responsibility, they also introduce unique security risks.
Convenience Meets Risk
AI browsers like Microsoft Edge’s Copilot and Perplexity’s Comet and others implementing AI features are becoming smarter. They promise to handle everything from shopping to managing emails, making our online lives easier than ever. However, this leap in convenience brings fresh risks. Unlike traditional browsers, these AI agents can autonomously click links, fill out forms, and even make purchases—sometimes without adequate skepticism or user oversight.
How AI Gets Fooled
Recent tests reveal that AI browsers can fall for classic scams such as fake online stores or phishing emails—mistakes that many human users have learned to avoi
Attackers are also now designing sophisticated tricks like “prompt injection,” where hidden instructions in websites can manipulate the AI into downloading files or leaking sensitive data.
This growing complexity in online scams—driven by automation and AI behavior—has been termed “Scamlexity” by researchers at Guardio.
Existing security tools, like Safe Browsing filters, aren’t always equipped to protect against these new threats. Since AI agents can automate tasks at scale, a single vulnerability could be exploited across thousands—or even millions—of users.
Security needs to be built into the core of AI browsers. As AI becomes more integrated into our daily routines, both users and developers must prioritize security to ensure that innovation doesn’t come at the cost of trust.
Leave a Reply