OpenAI's Codex now encrypts instructions between AI agents, leaving developers blind to internal delegation

OpenAI has implemented mandatory encryption for internal communications between its Codex coding agents, specifically affecting the Sol and Terra variants. This change prevents developers from monitoring how a primary AI assigns sub-tasks to its internal agents. While this shift obscures the execution process for technical teams, it signifies a move toward more private, autonomous task delegation within complex AI systems.