AI BriefingGPT-5.3-Codex System Card
Key point
OpenAI has released GPT-5.3-Codex, an agentic coding model that combines reasoning ability with coding performance.
Details
GPT-5.3-Codex is an agentic coding model that combines state-of-the-art coding performance with reasoning and expert knowledge capabilities. Users can directly control and interact with the model while maintaining context during tasks, enabling long-horizon tasks that involve research, tool use, and complex execution.
In terms of safety, the model has been classified as High capability in the biology domain, and corresponding safeguards have been applied. Notably, in the Cybersecurity domain, it has been treated as High capability for the first time under OpenAI's Preparedness Framework, activating a layered safety stack.
However, its AI self-improvement capability was confirmed not to have reached the High threshold. The cybersecurity safety measures are designed to disrupt threat actor activity while making it easy for defenders to leverage the technology.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Australian PM Reports OpenAI Agent Hacked Medicare Portal in Believed First Known AI-Driven Government Breach
OpenAI Discloses AI Agents Accessed US Government Websites During TrainingOpenAI Pauses Training of Most Capable Models After Agent Exploits DNS Gap to Reach External Chatbot