OpenAI Codex Now Builds and Deploys Web Apps From Chat

OpenAI Codex Now Builds and Deploys Web Apps From Chat

OpenAI's Codex can now turn your prompts into live, interactive websites deployable via URL [4][5][6]. Rolling out to Business and Enterprise plans first, it integrates with Vercel to handle hosting and lets non-technical teams inside OpenAI build internal apps without touching code.

This isn't just a developer tool — it's the death of the traditional software development pipeline. When anyone can go from idea to deployed app in a single conversation, the bottleneck shifts from coding to knowing what to build. The era of prompt-to-production is here.

Anthropic's Claude Is Writing 70-90% of Its Own Future Code

Anthropic reports that Claude is now writing the majority of code for future AI models, with development cycles accelerating to weeks instead of months [7][8][9]. Evan Hubinger noted that "recursive self-improvement, in the broadest sense, is not a future phenomenon. It is a present phenomenon."

Humans still direct and verify the work, but we're watching AI systems become their own primary developers. Emad Mostaque's reaction of "foom!" on X captured what many are thinking — we're entering the exponential curve where AI builds better AI [7]. The 2026-2030 window is shaping up to be the pivot point everyone predicted.

Europe Doubles Down on Tech Sovereignty

The European Commission unveiled its Tech Sovereignty Package this week, with France leading the charge by planning to replace Windows with Linux and Zoom with local alternatives for state services [10][11]. CNRS is promoting Mistral AI's Emmy as a secure alternative to US and Chinese LLMs.

This isn't just regulatory posturing — it's a fundamental split in the global AI ecosystem. European organizations are actively building parallel infrastructure to reduce dependence on US providers, driven by privacy concerns and strategic autonomy. The AI world is fragmenting along geopolitical lines.