OpenAI patched a command injection flaw in its Codex CLI tool that let attackers run arbitrary commands on developer machines ...
Sam Altman also called for staffers to delay progress on other efforts such as autonomous AI agents and advertising.