Addendum to GPT-5.2 System Card: GPT-5.2-Codex
OpenAI just shipped GPT-5.2-Codex with sandboxed agents and configurable network access. Here's what changed.
Why it matters
GPT-5.2-Codex represents a capability escalation in code generation with explicit safety architecture for agent deployment. For founders building on OpenAI models, this signals both new agentic capabilities and stricter guardrails around code execution and network access.
The key facts
9 to knowGPT-5.2-Codex released with specialized safety training for harmful tasks
Model-level mitigations include prompt injection defenses
Product-level mitigations: agent sandboxing and configurable network access
System card published Dec 18 2025 indicating formal safety documentation
Variant of GPT-5.2 line focused on code generation and agent workflows
GPT-5.2-Codex variant announced
Model-level mitigations: specialized safety training for harmful tasks and prompt injection defenses
Published December 18, 2025
System card format indicates pre-deployment safety disclosure
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: This system card outlines the comprehensive safety measures implemented for GPT‑5.2-Codex. It details both model-level mitigations, such as specialized safety training for harmful tasks and prompt injections, and product-level mitigations like agent sandboxing and configurable network access.