Just hours after ChatGPT-5’s launch, a GitHub user named elder-pilnius posted its system prompt—the hidden instruction that defines the AI’s role, tone, and constraints—as part of a hacker-activist effort for transparency. This leak raises security concerns because knowing the system prompt might let malicious actors manipulate the model’s behavior or bypass safeguards.

The incident highlights how revealing internal prompts could weaken AI defenses and reduce control over automated responses.