AI · 3h ago
OpenAI encrypts Codex agent instructions, sparking developer backlash
OpenAI has encrypted the internal messages of its MultiAgentV2 system, preventing developers from seeing how Codex agents make decisions. The move has drawn criticism for hindering debugging and auditing of AI agent behavior. Developers argue the encryption undermines transparency and trust in the platform.
Meridian48 take
OpenAI's encryption move prioritizes secrecy over developer needs, risking long-term trust for short-term competitive advantage.
Read the full reporting
OpenAI hides Codex agent instructions behind encryption, leaving developers in the dark →
The Register
openaicodex-agents