A significant data breach involving two AI companion applications, Chattee Chat and GiMe Chat, has exposed over 43 million private messages, highlighting the security vulnerabilities present when users entrust personal interactions to AI platforms.
On August 28, 2025, researchers at Cybernews identified a publicly exposed Kafka Broker server operated by Imagime Interactive Limited, the Hong Kong-based company behind the apps. The server was left without any security protection, meaning it had no authentication requirements or access controls, allowing anyone to access the data it contained. The server was actively streaming real-time conversations between users and their AI companions. The exposed data included not only text-based messages but also direct links to personal photos, videos, and AI-generated images exchanged within the apps. Researchers described some of the exposed content as “virtually not safe for work,” indicating the intimate and sensitive nature of the leaked information.
The breach affected a total of 400,000 users across both iOS and Android platforms. According to the investigation, approximately two-thirds of the exposed data originated from iOS users, with the remaining one-third coming from users on Android devices. The majority of the individuals affected by the leak were located in the United States.
While the leaked data did not include full names or email addresses, it did contain other significant identifiers, including user IP addresses and unique device IDs. This information can be cross-referenced with other data sources to track and potentially identify individuals. The analysis showed that users sent an average of 107 messages each to their AI partners, creating a substantial digital footprint for each user containing personal thoughts and interactions that could be leveraged for malicious purposes such as identity theft, targeted harassment, or blackmail.
The investigation also brought financial details to light. Purchase logs included in the exposed data revealed that some users spent considerable amounts of money on the apps, with individual spending reaching as high as $18,000 to interact with their AI companions. It is estimated that the developer had earned over $1 million in revenue from these applications before the data breach was discovered. In its privacy policy, Imagime Interactive Limited stated that user security was “of paramount importance.” However, the complete absence of authentication measures on the server directly contradicts this claim, revealing a critical failure to implement basic security safeguards for sensitive user data.
Upon discovering the vulnerability, Cybernews promptly reported the issue to Imagime Interactive Limited. The unsecured server was eventually taken offline in mid-September. Before its removal, the server had become listed on public IoT search engines, which are platforms that index internet-connected devices, making it easily discoverable for cybercriminals actively searching for vulnerable systems. It remains unclear whether any malicious actors accessed the compromised data before the server was secured. The potential for harm persists, as any downloaded conversations and images could still be used to facilitate sextortion scams, phishing attacks, or cause significant reputational damage to the affected users.
In response to the breach, cybersecurity experts outlined several tips for users to protect their data when using AI applications. Think before you share: Users should avoid sending personal or sensitive content through AI chat applications. Once data is shared, control over it is effectively lost. Use reputable AI tools: It is recommended to choose applications from developers with transparent privacy policies and a proven track record of strong security measures. Remove your data online: Employing a data removal service can help scrub personal information from public databases. While not a complete solution, it can limit the information available to scammers. Strengthen your cybersecurity with strong antivirus software: Installing reputable antivirus software provides a layer of defense by blocking scams, detecting intrusions, and alerting users to phishing attempts. Protect your accounts with a password manager and MFA: Using a password manager for strong, unique passwords and enabling multi-factor authentication (MFA) are critical steps to prevent unauthorized account access.
This data leak serves as a reminder that AI chat applications store vast quantities of highly sensitive data. When this data is compromised, it can lead to severe consequences, including blackmail, impersonation, and public embarrassment. The incident underscores the need for stronger security standards and greater accountability within the growing AI companion industry. For users, developing an awareness of how their data is handled and protected is a critical first step toward preventing personal information from being exposed online.
