AI Browsers: Perplexity's Comet Exposes New Web Security Risks

AI-powered web browsers are revolutionizing productivity, but they also introduce significant security risks. A recent vulnerability in Perplexity's Comet browser, known as "CometJacking," highlights these dangers, potentially exposing users to data theft and other cyber threats.

The CometJacking vulnerability, discovered by cybersecurity firm LayerX, allows malicious links to trick the browser's AI into executing harmful commands. This could lead to the extraction of sensitive data from users' emails and calendars, posing a serious risk to personal and professional information.

Attackers exploited this flaw by encoding stolen data, disguising it as harmless text before sending it to a remote server under their control. This sophisticated method bypassed Comet's defenses, underscoring the need for enhanced security measures in AI-powered browsers.

The Rise of AI Browsers and Emerging Risks

While Google Chrome remains the dominant player in the browser market, new AI-powered browsers like Comet are gaining traction. Experts predict that this could spark a new "browser war," as companies compete to offer the most advanced AI features.

However, the rush to integrate AI into browsers may introduce new vulnerabilities. Or Eshed, CEO of LayerX, warns that as AI features become more prevalent, browsing could become riskier. Both old and new attack methods may resurface, posing challenges for cybersecurity experts.

Perplexity's Response to the Vulnerability

LayerX reported the CometJacking vulnerability to Perplexity, but the company initially downplayed the issue. Perplexity later acknowledged the problem and claimed to have independently identified and patched it. A spokesperson for the company asserted that the vulnerability was never exploited and emphasized Perplexity's commitment to security and collaboration with the security community.

Industry Developments and Future Implications

In related news, chipmaker Advanced Micro Devices (AMD) secured a major deal with OpenAI, the creator of ChatGPT. OpenAI will acquire 6 gigawatts of AMD's AI chips over several years, signaling a move to diversify its chip supply and reduce reliance on Nvidia. AMD shares surged 25% following the announcement.

Additionally, OpenAI's Sora video-generation app is offering users more control over their likenesses in videos. Users can now specify how their images should or should not be used, addressing concerns about privacy and consent.

These developments highlight the rapidly evolving landscape of AI technology and the need for robust security measures to protect users from emerging threats.