OpenAI built an AI coding agent and uses it to improve the agent itself

Ars TechnicaFriday, December 12, 2025 at 10:16:42 PM
PositiveTechnology
OpenAI built an AI coding agent and uses it to improve the agent itself
  • OpenAI has developed an AI coding agent known as Codex, which is primarily responsible for its own enhancements, showcasing a self-improving technology. This innovation highlights OpenAI's commitment to advancing AI capabilities in coding and software engineering.
  • The introduction of Codex represents a significant step for OpenAI as it seeks to maintain its leadership in the competitive AI landscape, particularly in the face of growing competition from other tech giants like Google and Anthropic.
  • This development underscores a broader trend in the AI industry where companies are increasingly focusing on creating self-sustaining systems that can autonomously improve their performance, reflecting a shift towards more advanced and efficient AI solutions.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Continue Readings
OpenAI Ends ‘Vesting Cliff’ for New Employees in Compensation-Policy Change
PositiveTechnology
OpenAI has announced the removal of the 'vesting cliff' for new employees, a policy change aimed at easing restrictions that previously discouraged new hires from leaving the company. This adjustment comes amid a competitive landscape for talent, particularly in the AI sector, where companies like xAI are also vying for skilled workers.
OpenAI are quietly adopting skills, now available in ChatGPT and Codex CLI
NeutralTechnology
OpenAI has begun quietly adopting new skills that are now integrated into its ChatGPT and Codex CLI platforms, reflecting the company's ongoing commitment to enhancing user experience and functionality. This development is part of OpenAI's strategy to stay competitive in the rapidly evolving AI landscape.
ChatGPT’s adult mode is coming — and it might not be what you think it is
NeutralTechnology
OpenAI plans to introduce an adult mode for ChatGPT in 2026, contingent on the successful implementation of its age verification AI, which is currently in testing. This move aims to expand the capabilities of ChatGPT while addressing user safety and content appropriateness.
OpenAI signs deal to bring Disney characters to Sora and ChatGPT
NeutralTechnology
OpenAI has signed a significant deal with Disney to integrate its iconic characters into the Sora video generation tool and ChatGPT, following a $1 billion equity investment from Disney. This partnership aims to enhance the creative capabilities of Sora, allowing users to generate videos featuring beloved characters from Disney's extensive portfolio, including Marvel and Star Wars.
$1B for AI Slop? Why Disney Is Spending Big and Bringing Its Iconic Characters to OpenAI
PositiveTechnology
The Walt Disney Company has announced a substantial $1 billion investment in OpenAI, which includes a licensing agreement for over 200 iconic characters, such as Mickey Mouse and characters from Marvel, Star Wars, and Pixar, to be utilized on OpenAI's Sora video-making platform starting in early 2026.
I tested GPT-5.2 and the AI model's mixed results raise tough questions
NeutralTechnology
OpenAI has launched GPT-5.2, its latest AI model, which has shown mixed results in tests compared to its predecessor, GPT-5.1. The model was subjected to a series of text and image challenges, raising questions about the value for Plus subscribers.
Google’s Here with Yet Another Gemini Upgrade: It’s Deepest Research Agent Until Now
NeutralTechnology
Google has launched an upgraded version of its Gemini Deep Research agent, built on the Gemini 3 Pro reasoning model, coinciding with OpenAI's release of GPT-5.2. This strategic timing highlights a competitive landscape in AI technology, as both companies vie for dominance in advanced AI capabilities.
Third-party breach exposes ChatGPT account details
NegativeTechnology
A significant data breach involving ChatGPT has exposed personal information of users, including names and emails, through its partner Mixpanel. OpenAI has confirmed the security incident, raising alarms about user data safety.

Ready to build your own newsroom?

Subscribe to unlock a personalised feed, podcasts, newsletters, and notifications tailored to the topics you actually care about