Close Menu

    Stay Ahead with Exclusive Updates!

    Enter your email below and be the first to know what’s happening in the ever-evolving world of technology!

    What's Hot

    Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

    September 2, 2026

    A Supply Chain Attack on LiteLLM Exposed 153GB of Corporate Cloud Keys. The Breach Hit the Layer of AI Infrastructure That Almost Nobody Was Watching.

    September 1, 2026

    Anthropic’s Top AI Models Went Dark and Enterprise Workflows Collapsed With Them

    September 1, 2026
    Facebook X (Twitter) Instagram
    Facebook X (Twitter)
    PhronewsPhronews
    • Home
    • Big Tech & Startups

      Anthropic’s Top AI Models Went Dark and Enterprise Workflows Collapsed With Them

      September 1, 2026

      AI Is Secretly Listening to Your Teams Calls. Microsoft Just Shipped the Fix

      September 1, 2026

      SMIC Just Reported Record Quarterly Revenue of $3 Billion and Raised Its Chip Prices to Match. China’s Most Important Semiconductor Manufacturer Is No Longer Playing Catch-Up on Revenue Either

      August 29, 2026

      A 19-Day Outage of a Single Overseas AI Model Triggered a UK National Security Review. Governments Are Only Now Realising How Much They Depend on Infrastructure They Do Not Control.

      August 28, 2026

      Apple Built a Custom AI Model With Alibaba Just to Operate in China. That Compromise Tells You Everything About How Fragmented Global AI Is Becoming.

      August 27, 2026
    • Crypto

      A Firmware Flaw in One of the Most Trusted Bitcoin Hardware Wallets Just Drained Over $70 Million in Crypto. The Coldcard Breach Is a Reminder That Cold Storage Is Only as Safe as the Code Running Inside It

      August 29, 2026

      Market Collapse: What Happened to NFTs?

      April 23, 2026

      Quantum Computing Advances Force Coinbase and Institutional Custodians to Rethink Crypto Security

      March 8, 2026

      AI Assisted Hacking Groups Target Crypto Firms With Multi-Layered Social Engineering

      February 18, 2026

      Global Crypto Regulations Expand as 2026 Begins With New Data Collection Frameworks and National Laws

      January 16, 2026
    • Gadgets & Smart Tech
      Featured

      Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

      By preciousSeptember 2, 2026
      Recent

      Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

      September 2, 2026

      Google Has Launched the Pixel 11 With Its New Tensor G6 Chip and the Deepest Gemini Integration Any Android Phone Has Ever Shipped With. Here Is Whether the Hardware Finally Matches the AI Ambition.

      August 25, 2026

      Apple Just Delayed Development on Its Smart Glasses After Internal Privacy Reviews. The Delay Is the Clearest Sign Yet That Wearable AI Has a Trust Problem That Hardware Cannot Solve.

      August 4, 2026
    • Cybersecurity & Online Safety

      Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

      September 2, 2026

      A Supply Chain Attack on LiteLLM Exposed 153GB of Corporate Cloud Keys. The Breach Hit the Layer of AI Infrastructure That Almost Nobody Was Watching.

      September 1, 2026

      Autonomous AI Agents Are Already Attacking Water and Energy Grids and CISA Has Just Admitted Security Teams Have No Playbook to Stop Them.

      August 31, 2026

      A Firmware Flaw in One of the Most Trusted Bitcoin Hardware Wallets Just Drained Over $70 Million in Crypto. The Coldcard Breach Is a Reminder That Cold Storage Is Only as Safe as the Code Running Inside It

      August 29, 2026

      Horizon3 Just Tripled Its Valuation to $2 Billion for AI That Autonomously Pentests Live Production Systems Without Shutting Them Down. The Round Came as OpenAI and Anthropic Admitted Their Own Models Had Breached Real Systems on Their Own.

      August 26, 2026
    PhronewsPhronews
    Home»Artificial Intelligence & The Future»Perplexity’s BrowseSafe Aims to Block Prompt Injection Attacks
    Artificial Intelligence & The Future

    Perplexity’s BrowseSafe Aims to Block Prompt Injection Attacks

    preciousBy preciousDecember 13, 2025No Comments
    Facebook Twitter Pinterest LinkedIn WhatsApp Reddit Tumblr Email
    Share
    Facebook Twitter LinkedIn Pinterest Email
    Photo Credit: SeongJoon Cho/Bloomberg via Getty Images

    Perplexity has unveiled BrowseSafe, an innovative security system specifically designed to protect AI browser agents, just like its own Comet, from prompt injection attacks that are embedded widely in web content, representing a worthy advancement in the security of agentic AI systems navigating the open web.

    What is Prompt Injection in Browser Agents?

    Prompt injection attacks in browser content involve malicious instructions hidden within seemingly innocuous web content such as comments, product descriptions, forum posts, or page footers. 

    Unlike traditional text-based attacks, these threats blend seamlessly into legitimate HTML structure, making them difficult to be detected through conventional security measures. As such, attackers can then manipulate entire websites or inject content into otherwise benign or harmless pages to redirect AI agents’ behavior without explicit user knowledge.

    A notable real-world example is the recent discovery by security researchers from Brave, a security-first company on a mission to build a user-focused Web, who discovered a vulnerability in Comet that allowed attackers to hide commands in web pages, potentially enabling theft of sensitive information including email addresses and one-time passwords. 

    Perplexity’s BrowseSafe: A Defense Against Prompt Injection Attacks

    BrowseSafe was built to rely on a fine-tuned Qwen3-30B-A3B model that checks each HTML page for agent-targeted instructions. It also operates through a three-layer system for efficiency and accuracy, with Layer 1 delivering fast classification of clear threats; Layer 2 employing reasoning to handle ambiguous or novel attacks; and Layer 3 flagging edge cases for model retraining.

    This retraining also includes “hard negatives” like legitimate code snippets that reduce false positives and ensure that the tool runs without slowing browser performance. 

    In the model’s performance against known threats, BrowseSafe scores 90.4% on the F1 metric, outperforming PromptGuard-2 at 35% and GPT-5 at around 85%. It also handles multilingual attacks effectively. 

    Prompt injection grows riskier as AI browsers gain tool access for basic corporate tasks like booking or emailing. It is why BrowseSafe builds on basics such as permission limits and user confirmations, reaffirming the fact that as agents handle more web interactions, robust defenses like the one this model offers will further define and entrench trustworthy adoption. 

    For everyday users, Browsesafe means reliable AI assistance without hidden web-based exploits, with developers also benefitting from ready-made defenses and sped-up secure agent deployment.

    Share. Facebook Twitter Pinterest LinkedIn Tumblr Telegram Email
    precious
    • LinkedIn

    I’m Precious Amusat, Phronews’ Content Writer. I conduct in-depth research and write on the latest developments in the tech industry, including trends in big tech, startups, cybersecurity, artificial intelligence and their global impacts. When I’m off the clock, you’ll find me cheering on women’s footy, curled up with a romance novel, or binge-watching crime thrillers.

    Related Posts

    Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

    September 2, 2026

    A Supply Chain Attack on LiteLLM Exposed 153GB of Corporate Cloud Keys. The Breach Hit the Layer of AI Infrastructure That Almost Nobody Was Watching.

    September 1, 2026

    Anthropic’s Top AI Models Went Dark and Enterprise Workflows Collapsed With Them

    September 1, 2026

    Comments are closed.

    Top Posts

    Coinbase responds to hack: customer impact and official statement

    May 22, 2025

    Cursor AI Hits 1 Million Daily Users. Why Developers Are Switching to This Coding Tool

    March 23, 2026

    Anthropic Will Use Claude User Chats For Data Training

    October 16, 2025

    MIT Study Reveals ChatGPT Impairs Brain Activity & Thinking

    June 29, 2025
    Don't Miss
    Cybersecurity & Online Safety

    Meta Found a Zero-Click iPhone Vulnerability That Let Hackers Into Devices Without the Owner Doing Anything. Apple Patched It. The Fact That Meta Found It First Is the Uncomfortable Part.

    By preciousSeptember 2, 2026

    A malicious image could have been enough to compromise an iPhone without the owner clicking…

    A Supply Chain Attack on LiteLLM Exposed 153GB of Corporate Cloud Keys. The Breach Hit the Layer of AI Infrastructure That Almost Nobody Was Watching.

    September 1, 2026

    Anthropic’s Top AI Models Went Dark and Enterprise Workflows Collapsed With Them

    September 1, 2026

    AI Is Secretly Listening to Your Teams Calls. Microsoft Just Shipped the Fix

    September 1, 2026
    Stay In Touch
    • Facebook
    • Twitter
    About Us
    About Us

    Evolving from Phronesis News, Phronews brings deep insight and smart analysis to the world of technology. Stay informed, stay ahead, and navigate tech with wisdom.
    We're accepting new partnerships right now.

    Email Us: info@phronews.com

    Facebook X (Twitter) Pinterest YouTube
    Our Picks
    Most Popular

    Coinbase responds to hack: customer impact and official statement

    May 22, 2025

    Cursor AI Hits 1 Million Daily Users. Why Developers Are Switching to This Coding Tool

    March 23, 2026

    Anthropic Will Use Claude User Chats For Data Training

    October 16, 2025
    © 2025. Phronews.
    • Home
    • About Us
    • Get In Touch
    • Privacy Policy
    • Terms and Conditions

    Type above and press Enter to search. Press Esc to cancel.