—
Source: Fortune.com
Summary
OpenAI has released its new AI-powered coding model, GPT-5.3-Codex, which shows higher performance on coding benchmarks than rival models. However, the company is delaying full access and imposing tight controls due to serious cybersecurity concerns. The model’s capabilities raise the risk of automating cyber attacks, and OpenAI is taking a precautionary approach to mitigate these risks.
Our Reading
The announcement sounds familiar.
OpenAI is releasing a powerful new coding model, but it’s delayed full access due to cybersecurity concerns. The model’s capabilities could enable real-world cyber harm if automated or used at scale.
The company is rolling out the model with tight controls and a new trusted-access program for vetted security professionals. CEO Sam Altman said the model hits “high” for cybersecurity on OpenAI’s preparedness framework.
This is the first model OpenAI believes is good enough at coding and reasoning that it could meaningfully enable real-world cyber harm.
OpenAI’s caution reflects the reality that the same capabilities that make GPT-5.3-Codex effective at writing, testing, and reasoning about code also raise serious cybersecurity concerns.
Original Observation: This is OpenAI’s “high-risk” model.
Author: Evan Null
