Bugs Found in Perplexity AI's Chatbot Android App: A Deep Dive into Security Vulnerabilities

As artificial intelligence continues to advance, one of the most widely recognized applications in the field is the AI chatbot, Perplexity. Launched in late 2022, it quickly gained attention for its ability to provide highly accurate and well-researched responses, setting itself apart from competitors. However, this innovation has come with a cost—security. Recent findings by Appknox researchers have revealed that the Android version of Perplexity’s chatbot app is riddled with vulnerabilities that could jeopardize user security. In this article, we’ll explore the critical issues identified and analyze what they mean for both users and the company behind the app.

Security Concerns in Perplexity

Researchers from Appknox have uncovered 11 significant security bugs within the Perplexity AI Android app, raising concerns about its robustness in the face of potential attacks. While Perplexity AI was hailed for its accuracy and user-oriented features, these vulnerabilities expose a serious gap in its mobile security infrastructure.

The security audit, performed in early April 2025, revealed a worrying number of issues, some of which are common across other AI chatbots, such as DeepSeek. However, Perplexity’s app presented several unique risks. Below is a rundown of the key findings:

Insecure Network Configurations: These settings increase the likelihood of network-based attacks, though the severity was not assigned a specific Common Vulnerability Scoring System (CVSS) rating.
Absence of SSL Validation or Certificate Pinning (CVSS 5.9): Without proper server authentication, attackers could potentially launch impersonation attacks.
Weak Rooting/Jailbreak Detection (CVSS 6.8): Attackers could gain escalated privileges by bypassing device root detection.
StrandHogg Vulnerability (CVSS 6.5): This vulnerability in Android’s task management system allows attackers to hijack the app.
Exposure to CVE-2017-13156 (CVSS 6.7): This flaw permits attackers to modify Android apps without invalidating their signatures.
Clickjacking (CVSS 4.8): The app’s interface could be manipulated to trigger unintended actions on the user’s device.

Beyond these shared vulnerabilities, researchers discovered issues unique to Perplexity:

Cross-Origin Resource Sharing (CORS) Misconfigurations: This flaw in API responses allows malicious sites to communicate with Perplexity’s backend.
No SSL Pinning Implementation: This leaves the app open to man-in-the-middle (MitM) attacks.
Bytecode Lacking Obfuscation: The lack of obfuscation makes the app’s code vulnerable to reverse engineering.
No Detection for ADB or Developer Options: Attackers could exploit developer tools to manipulate the app in a controlled environment.

But perhaps the most concerning discovery was the presence of hardcoded secrets in the app’s code.

The Most Critical Vulnerability: Hardcoded Secrets

The most alarming vulnerability found in Perplexity’s Android app was the hardcoding of sensitive information, including Google API keys and access tokens. These credentials could allow malicious actors to bypass authentication layers and directly access the app’s APIs. This could result in severe data breaches, including the compromise of user data and overall app integrity.

According to Raghunandan J, head of research at Appknox, the discovery of hardcoded secrets is a red flag. Attackers with access to these secrets could effectively bypass security measures, causing significant damage. “If an attacker could get their hands on this, they could easily bypass all the authentication layers,” said J. This is a common issue with mobile apps, particularly those rushed to market, where security may take a backseat to speed.

Until these vulnerabilities are addressed, Appknox has strongly recommended that Android users uninstall the app to mitigate the risks associated with these security flaws.

What Undercode Says:

The discovery of security vulnerabilities in Perplexity AI’s Android app highlights a critical issue facing many companies in the fast-moving AI and mobile app development space. The rush to release new applications often results in a lack of focus on proper security measures. This is particularly evident in Perplexity’s case, where issues that were already identified in another chatbot, DeepSeek, appear to have been ignored or overlooked.

For a chatbot with such high expectations, especially given its early market presence, these vulnerabilities could significantly harm its reputation. The lack of SSL pinning and the misconfigurations in CORS can lead to easily exploitable scenarios for cybercriminals. Not only does this jeopardize user data, but it also exposes the company to potential regulatory and legal challenges.

Security is no longer a luxury but a necessity, especially for apps handling sensitive information. With AI-driven services becoming ubiquitous, users expect nothing less than rigorous security measures. The absence of such measures in the Perplexity app suggests that the company may have prioritized features over safeguarding its users.

In addition, the fact that hardcoded API keys were found exposes another layer of negligence, which is particularly concerning in apps that deal with personal data. These secrets are often a prime target for attackers, and their inclusion in a production app can significantly increase the chances of a breach.

The broader issue here is a failure to integrate security into the development process from the very beginning. In the race to deliver new features or products, security often gets sidelined. This case serves as a wake-up call for developers, emphasizing the importance of conducting thorough security assessments before launching a product. Additionally, developers need to prioritize secure coding practices, such as using secure storage mechanisms for sensitive information and ensuring that network communications are encrypted and authenticated.

As AI chatbots like Perplexity become increasingly integrated into daily life, the importance of robust security cannot be overstated. The risks posed by these vulnerabilities are too great to ignore, and it is imperative that the company takes swift action to address them before they can cause significant harm.

Fact Checker Results:

Security Risks Validated: The vulnerabilities identified by Appknox, including insecure network configurations and hardcoded secrets, align with common security flaws in mobile applications.
Potential for Data Breach: Hardcoded API keys are indeed a significant risk, and their exposure could lead to data theft or compromise.
Developer Negligence: The issues found suggest that security was not prioritized during development, which is consistent with the findings from other rushed-to-market applications.