282 iOS AI Apps Leak API Keys and Open AI Proxy Access in Network Traffic Study

A study of 444 AI chatbot apps for iPhone revealed that 282 of them (nearly two-thirds) exposed paid AI access through their network traffic. The exposure was discovered by researchers who analyzed the apps' network communications and found plaintext API keys, reusable tokens, or backend servers that accepted requests with no authentication. This allowed anyone who intercepted the traffic to send model requests on the developer's account, potentially incurring costs or accessing proprietary AI services.

The attack vector was a misconfiguration in the apps' network security, where sensitive credentials were transmitted in plaintext or not required at all. The exposed data types included API keys and reusable tokens, which could be used to access large language model (LLM) services such as OpenAI's proxy. No specific threat actor was attributed, and no CVEs or MITRE ATT&CK techniques were mentioned in the article.

No post-incident developments were reported, such as regulatory actions, litigation, or remediation milestones. The study was publicly disclosed in June 2026, but the exact discovery date by the researchers was not specified.