Hey Devs! This week in AI marks another shift in how major players are shaping the ecosystem, from OpenAI’s new advertising strategy aimed at expanding access to Google’s big push into agentic commerce with UCP, to fresh generative model and developer tool releases. Let’s jump into the most important updates.
OpenAI Adds Ads to ChatGPT
OpenAI will soon show ads in ChatGPT to free users and to new ChatGPT Go plan subscribers. Ads will appear below responses and are clearly labeled. They won’t change the AI’s answers or share your private chat data with advertisers. Paid plans (Plus, Pro, Business, Enterprise) stay ad-free.
Read more: https://openai.com/index/our-approach-to-advertising-and-expanding-access/
Universal Commerce Protocol (UCP) - Open Shopping Standard
Google and partners, including Shopify, Walmart, and others, launched the Universal Commerce Protocol (UCP). It’s an open standard that lets AI assistants (like Gemini and others) talk to store systems to search for products, add items to the cart, and check out automatically. It removes custom integrations, making agent-based shopping easier to build.
Read more: https://developers.googleblog.com/under-the-hood-universal-commerce-protocol-ucp/
TranslateGemma - New Open Translation Models
Google released TranslateGemma, a set of open translation models built on Gemini 3. They come in 4B, 12B, and 27B sizes, support 55 languages, and can even translate text inside images. The 12B model performs very well, sometimes better than the 27B one. Models and data are available on usual open model sites for developers to use.
Read more: https://blog.google/innovation-and-ai/technology/developers-tools/translategemma/
Veo 3.1 - Better Generative Video
Google updated its Veo video model to version 3.1 with mobile-focused features:
Vertical (9:16) video
HD/4K output
Better character dialogue and style
This makes AI video generation much more useful for apps and content creation.
Read more: https://blog.google/innovation-and-ai/technology/developers-tools/enhanced-veo-3-1-gemini
GitHub Copilot CLI Gets Smarter
GitHub updated Copilot CLI with new models like GPT-5 mini and GPT-4.1, built in free for users. It also added automation helpers (like Explore, Task, Plan, and Code-review), plus better context tools for scripting and CI. This boosts productivity for coding and dev workflows.
Codex Skills Explained: Stop Repeating Prompts, use SKILLS
Codex Skills are one of the most powerful and underrated features in OpenAI Codex. Instead of repeating prompts, skills let you package workflows into reusable, discoverable instructions that Codex loads only when needed.
Watch on YouTube:
Thanks for reading! If you enjoyed this round-up, hit subscribe and share any thoughts or news in the comments. See you next week with more AI dev updates!
Cheers! 😉