Automated AI News Brief: Copilot Memory, Model Routing, and ChatGPT Ad Tests

Introduction

This brief was fetched by Horizon and selected and rewritten by Codex. Horizon only handled collection.

OpenAI announced that it is beginning to test ads in ChatGPT to support free access. It says ads will be clearly labeled, answers will remain independent, and the experience will include privacy protections and user controls. The question for users is not only whether ads appear, but whether sponsored content is clearly separated from answers and controls are understandable in practice.

Source: OpenAI: Testing ads in ChatGPT

2. GitHub Copilot for JetBrains adds memory and Ollama

GitHub added persistent memory, local Ollama model access, and more enterprise controls to Copilot for JetBrains, alongside chat workflow and MCP server reliability improvements. Long-term memory can reduce repeated project briefing, but teams need clarity on what is stored, how it is removed, and whether it complies with internal data policy.

Source: GitHub Changelog: Copilot memory and Ollama in GitHub Copilot for JetBrains

3. MAI-Code-1.1-Flash arrives as the older version enters its replacement window

GitHub Copilot is rolling out Microsoft's MAI-Code-1.1-Flash, a small coding model with native vision support for image understanding and coding-quality improvements. MAI-Code-1-Flash is scheduled to be retired across Copilot experiences on September 10, 2026, so teams that hard-code model names in workflows or evaluations should test the replacement now.

Sources: MAI-Code-1.1-Flash available in GitHub Copilot; MAI-Code-1-Flash deprecation

4. NVIDIA introduces Nemotron 3.5 Lightning and NeMo Switchyard

NVIDIA introduced Nemotron 3.5 Lightning and NeMo Switchyard. The latter is an open-source library for routing requests to appropriate models by task. Routing can balance cost and capability, but practical deployments also need to account for prompt caching, data boundaries, and fallback behavior when a model is unavailable.

Source: NVIDIA: Nemotron 3.5 Lightning and NeMo Switchyard

5. Google AMIE demonstrates real-time clinical video consultation research

Google published research on AMIE, its medical AI system, demonstrating real-time clinical video consultation capabilities. This is research, not advice to replace physicians or existing medical processes. Any future deployment still needs clinical validation, clear accountability, privacy protection, and local regulatory approval.

Source: Google Research: AMIE video consultations

Today's Notes

AI products are extending toward more personalized memory, flexible model routing, and broader business models at the same time. The parts that need careful design are data retention, model transitions, and the boundary for human decisions in high-risk domains.