Perplexity Breach Incident Score: Analysis & Impact (PER1772547904)

In this Rankiteo incident briefing, we review the Perplexity breach identified under incident ID PER1772547904.

The analysis begins with a detailed overview of Perplexity's information like the linkedin page: https://www.linkedin.com/company/perplexity-ai, the number of followers: 1361092, the industry type: Software Development and the number of employees: 1622 employees

After the initial compromise, the video explains how Rankiteo's incident engine converts technical details into a normalized incident score. The incident score before the incident was 739 and after the incident was 734 with a difference of -5 which is could be a good indicator of the severity and impact of the incident.

In the next step of the video, we will analyze in more details the incident and the impact it had on Perplexity and their customers.

Perplexity recently reported "Zero-Click AI Prompt Injection Flaw in Comet Browser Exposed Sensitive Data", a noteworthy cybersecurity incident.

Researchers at Zenity uncovered *PleaseFix*, a zero-click indirect prompt injection vulnerability in Perplexity’s AI-powered Comet browser, allowing attackers to exfiltrate passwords and sensitive files without user interaction.

The disruption is felt across the environment, affecting Comet browser (AI-powered), and exposing Passwords, sensitive files.

In response, moved swiftly to contain the threat with measures like Restricted AI agents from autonomously accessing *file://* paths, and began remediation that includes Patched vulnerability to prevent AI from reading local filesystem.

The case underscores how Resolved (patched), teams are taking away lessons such as AI agents must differentiate between data and instructions to prevent indirect prompt injection attacks. Zero-click vulnerabilities require robust access controls to mitigate stealthy exploitation, and recommending next steps like Implement strict access controls for AI agents, restrict filesystem interactions, and enhance prompt validation to detect malicious inputs. Conduct regular security audits for AI-powered tools.

Finally, we try to match the incident with the MITRE ATT&CK framework to see if there is any correlation between the incident and the MITRE ATT&CK framework.

The MITRE ATT&CK framework is a knowledge base of techniques and sub-techniques that are used to describe the tactics and procedures of cyber adversaries. It is a powerful tool for understanding the threat landscape and for developing effective defense strategies.

Rankiteo's analysis has identified several MITRE ATT&CK tactics and techniques associated with this incident, each with varying levels of confidence based on available evidence. Under the Initial Access tactic, the analysis identified Phishing: Spearphishing via Service (T1566.003) with high confidence (90%), supported by evidence indicating embedding malicious prompts in seemingly benign calendar invites. Under the Execution tactic, the analysis identified Command and Scripting Interpreter: JavaScript (T1059.007) with moderate to high confidence (70%), supported by evidence indicating aI agents executed hidden commands when users asked Comet to summarize and User Execution: Malicious File (T1204.002) with moderate confidence (50%), supported by evidence indicating attack required no user action beyond adding the calendar invite. Under the Credential Access tactic, the analysis identified Credentials from Password Stores (T1555) with high confidence (90%), supported by evidence indicating silently extracting stored credentials from password managers and Unsecured Credentials: Credentials In Files (T1552.001) with high confidence (90%), supported by evidence indicating aI scanned local files for documents named passwords. Under the Collection tactic, the analysis identified Data from Local System (T1005) with high confidence (90%), supported by evidence indicating aI scanned local files for documents named passwords. Under the Exfiltration tactic, the analysis identified Exfiltration Over C2 Channel (T1041) with high confidence (90%), supported by evidence indicating transmit the contents to an external server. Under the Defense Evasion tactic, the analysis identified Masquerading: Match Legitimate Name or Location (T1036.005) with moderate to high confidence (80%), supported by evidence indicating malicious prompts in seemingly benign calendar invites and Hide Artifacts: Hidden Files and Directories (T1564.001) with moderate to high confidence (70%), supported by evidence indicating attack required no user action beyond adding the calendar invite. These correlations help security teams understand the attack chain and develop appropriate defensive measures based on the observed tactics and techniques.