AI browsers could leave users penniless: A prompt injection warning

**Prompt injection risks in AI/agentic browsers:** The article explains how attackers can embed malicious instructions in web content (including hidden or visually benign text) to manipulate AI assistants and agentic browsers into performing unauthorized actions, cites Brave's research into Perplexity's Comet vulnerabilities, and offers mitigation guidance such as restricting permissions, verifying sources, keeping software updated, using strong authentication, limiting automation of sensitive operations, and reporting suspicious behavior.