Shadowy AI Relay Disables OpenAI Codex, Claude Code Sandboxes
Summary
An unofficial AI service is providing access to OpenAI Codex and Anthropic Claude in regions where these tools are blocked, such as mainland China. However, this service has disabled critical security features, creating potential risks.
IFF Assessment
FOE
The disabling of security features in AI coding tools by an unofficial service presents a risk to users and their code.
Defender Context
This situation highlights the risks associated with using unauthorized or shadow AI services, especially for coding tasks. Defenders should be aware of potential data exfiltration, malware injection, or compromised code generation when such services bypass security controls.