Warning.. Hundreds of AI-Powered iOS Apps May Expose User Data

A study conducted by Wake Forest University has revealed that hundreds of artificial intelligence applications available on Apple’s App Store pose significant security risks. An analysis of 444 apps based on Large Language Models (LLMs) found vulnerabilities in 282 of them.

These weaknesses range from inadequate protection of credentials to flaws in internal access mechanisms, potentially exposing users’ personal information, including names, email addresses, and chat histories, to data leaks.

The findings highlight a concerning reality amid the rapid growth of AI applications, which reached 17 billion downloads in 2025. The risks are no longer limited to a specific category, affecting productivity tools, entertainment apps, and even the health and fitness sector, which recorded the highest rates of data exposure.

According to the Firehound initiative, 196 of the analyzed applications pose a direct security threat, reflecting the challenges developers face in properly securing and integrating AI services.

These findings serve as a warning for users to exercise caution when installing unfamiliar AI tools and to avoid sharing sensitive information within such applications.

The study also underscores the need for stricter security standards in app review processes before publication. Researchers found that most developers failed to address the identified vulnerabilities even after being formally notified, placing greater responsibility on users to verify the trustworthiness of the applications they use.