OpenAI released GPT-5.3-Codex on February 5, 2026, marking the first publicly released model the company says contributed to its own development process. The model is optimized for agentic coding tasks with enhanced autonomous code generation capabilities. OpenAI describes the recursive involvement as a milestone in AI self-improvement, though specific details on the extent of the model's contribution to its own training remain limited.
AI coding tools have enabled a flood of bad code that threatens to overwhelm many projects. Building new features is easier, but maintaining them is just as hard.
Called Automations, the new system gives users a way to automatically launch agents within their coding environment, triggered by a new addition to the codebase, a Slack message, or a simple timer.
Google DeepMind's AlphaEvolve, a Gemini-based coding agent, has recovered 0.7% of Google's total compute budget through automated optimization. The system also independently discovered novel mathematical structures, marking a departure from AI as a tool-assistant toward AI as a research contributor. The announcement was made March 6, 2026.
Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv and feedback from readers. If you’d like to support this, please subscribe. Subscribe now AI progress is moving faster than even well regarded forecasters can guess:…Ajeya Cotra updates her timelines…“On Jan 14th, I made p
Anthropic launched Code Review in Claude Code, a multi-agent system that automatically analyzes AI-generated code, flags logic errors, and helps enterprise developers manage the growing volume of code produced with AI.
Even seemingly simple engineering tasks — like updating an API — can become monumental undertakings when you’re dealing with millions of lines of code and thousands of engineers, especially if the changes are security-related. Nowhere is this more apparent than in mobile security, where a single cla
Anthropic’s new auto mode for Claude Code lets AI execute tasks with fewer approvals, reflecting a broader shift toward more autonomous tools that balance speed with safety through built-in safeguards.
Microsoft added Critique and Council features to Copilot on March 30, 2026, enabling GPT and Claude to operate sequentially or in parallel within the same workflow. Multi-model configurations outperformed single-model baselines on internal benchmarks. The move makes multi-LLM orchestration a native Microsoft product feature rather than a developer pattern.
I used coding agents to build agents that automated part of my job. Here's what I learned about working better with coding agents. The post Agent-driven development in Copilot Applied Science appeared first on The GitHub Blog.
/fleet lets Copilot CLI dispatch multiple agents in parallel. Learn how to write prompts that split work across files, declare dependencies, and avoid common pitfalls. The post Run multiple agents at once with /fleet in Copilot CLI appeared first on The GitHub Blog.
Discover how Rubber Duck provides a different perspective to GitHub Copilot CLI. The post GitHub Copilot CLI combines model families for a second opinion appeared first on The GitHub Blog.
A survey carried out by OutSystems, The State of AI Development 2026 [email wall], argues that AI has moved into early production phase for many enterprises, primarily inside the IT function. The survey was based on the responses of 1,879 IT leaders, and warns that adoption of AI is in danger of run
Linus Torvalds and maintainers just finalized the Linux kernel's new AI policy - but it might not address the biggest challenge with AI-generated code. Here's why.
Fordel Studios