# Morning AI TL;DR — 2026-06-14

Freshness note: the last 24–48 hours were relatively light on concrete model/tool launches, so today mixes the strongest fresh policy/platform items with a few still-relevant developer releases from the past ~10 days.

- **GitHub Copilot code review added more enterprise controls** — GitHub’s June 12 changelog says code review now supports organization runner controls, Copilot content exclusion, and no character limit on repository custom instructions, making AI review easier to govern at scale. ([GitHub Changelog](https://github.blog/changelog/), [Microsoft Developer Changelog](https://developer.microsoft.com/en-us/changelog))
- **Canada moved to regulate AI chatbots after a school shooting** — Reuters reports new legislation is already drawing scrutiny over loopholes and long implementation timelines, a reminder that child-safety and chatbot-liability rules are moving from theory into product risk. ([Reuters](https://www.reuters.com/legal/litigation/canadas-move-rein-ai-chatbots-spurred-by-school-shooting-faces-doubts-over-2026-06-12/))
- **OpenAI faces a coalition state-AG investigation** — Reuters/CNBC report U.S. state attorneys general are investigating OpenAI, with OpenAI saying it will engage constructively; builders should expect more scrutiny around consumer safety, transparency, and governance. ([Reuters](https://www.reuters.com/business/openai-under-investigation-by-coalition-state-attorneys-general-wsj-reports-2026-06-12/), [CNBC](https://www.cnbc.com/2026/06/12/openai-says-its-engaging-constructively-with-state-ags-.html))
- **Salesforce published Claude Code “Salesforce Skills” guidance** — The June 9 developer post shows natural-language generation of Apex, LWCs, and tests, pointing to a practical path for Salesforce teams to encode platform conventions into coding-agent workflows. ([Salesforce Developers](https://developer.salesforce.com/blogs/2026/06/build-production-ready-apps-in-claude-code-with-salesforce-skills))
- **Google’s Gemma 4 12B is a local multimodal open model to watch** — Google says the 12B model can run on 16GB-class machines and natively handle text, image, audio, and video, which is useful for private/offline agent prototypes and edge workflows. ([Google](https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12B/), [Google Developers Blog](https://developers.googleblog.com/gemma-4-12b-the-developer-guide/))
- **Gemini CLI/API migration pressure continues** — Google’s Gemini tooling notes and migration chatter indicate developers should verify which Gemini CLI/API paths survive June changes before baking them into automation. ([Gemini CLI release notes](https://geminicli.com/docs/changelogs/), [Gemini API changelog](https://ai.google.dev/gemini-api/docs/changelog))

**Worth watching:** Enterprise AI work is shifting toward “operational resilience”: policy-aware model access, auditable coding agents, and vendor/toolchain fallback plans will matter as much as raw model quality.