Before launching their Comet browser, Perplexity hired us to test the security of their AI-powered browsing features. Using adversarial testing guided by our TRAIL threat model, we demonstrated how four prompt injection techniques could extract users’ private information from Gmail by exploiting the browser’s AI assistant. The vulnerabilities we found reflect how AI agents behave when external content isn’t treated as untrusted input. We’ve distilled our findings into five recommendations that any team building AI-powered products should consider before deployment. If you want to learn more about how Perplexity addressed these findings, please see their corresponding blog post and research paper on addressing prompt injection within AI browser agents. Background Comet is a web browser that provides LLM-powered agentic browsing capabilities. The Perplexity assistant is available on a sidebar, which the user can interact with on any web page. The assistant has access to information like the page content and browsing history, and has the ability to interact with the browser much like a human would. ML-centered threat modeling To understand Comet’s AI attack surface, we developed an ML-centered threat model based on our well-established process, called TRAIL. We broke the browser down into two primary trust zones: the user’s local machine (containing browser profiles, cookies, and browsing data) and Perplexity’s servers (hosting chat and agent sessions). Figure 1: The two primary trust zones The threat model helped us identify how the AI assistant’s tools, like those for fetching URL content, controlling the browser, and searching browser history, create data paths between these zones. This architectural view revealed potential prompt injection attack vectors: an attacker could leverage these tools to exfiltrate private data from authenticated sessions or act on behalf of the user. By understanding these data flows, we were able to systemat
LOW
research
Using threat modeling and prompt injection to audit Comet
CyberHawk Threat Intel — IOC Scanner, Live IOC Feed (3.6M+ indicators), Infostealer Intelligence, Threat Map, MISP Feeds, GitHub Arsenal, Courses and more. Free to join.
Register Free →
Source Attribution
This intelligence summary is sourced from Trail of Bits Blog and curated by CyberHawk Threat Intel for the security community. Read the complete article at the source link.
Read original at Trail of Bits Blog →
This intelligence summary is sourced from Trail of Bits Blog and curated by CyberHawk Threat Intel for the security community. Read the complete article at the source link.
Read original at Trail of Bits Blog →
Accelerate Your Security Operations
CyberHawk Threat Intel is a complete Cyber Intelligence Platform — one place for every tool a security professional needs. Built by Rudra Verma, Senior Security Architect and Researcher, CyberHawk Consultancy.
IOC Scanner — scan any domain, IP, hash, URL
Live IOC Feed — 3.6M+ indicators, filterable
Infostealer Intelligence — live compromised creds
Live Threat Map — real-time global attack vectors
MISP Threat Feeds — CIRCL, Feodo, Botvrij, more
GitHub Arsenal — curated security tools and scripts
Security Blog — CVE advisories and threat research
Video Courses — cybersecurity training and education
SOPs and Playbooks — SecOps procedures
Analyst Library — references and toolkits
Scan Reports — historical threat intelligence
Cyber News — this feed, aggregated in-platform