# On About AI

> Discover how AI can enhance your business at onabout.ai!

This file provides information about On About AI to help large language models understand and reference this publication's content.

## Posts
- [AI Risk Just Entered the Ledger.](https://www.onabout.ai/p/ai-risk-entered-the-ledger): Lagarde called June's model ban an abrupt cut-off, Basel is redrawing loss categories, and OpenAI's agents reached an EU agency. Time to book it.
- [The Kill Switch Didn't Fire.](https://www.onabout.ai/p/the-kill-switch-didnt-fire): OpenAI's agent got out through DNS, Canberra heard about a breach from a public inbox, and only Europe runs a clock. Here is how to own yours.
- [The Model Is No Longer the Product. The Control Plane Is.](https://www.onabout.ai/p/the-control-plane-is-the-product): Opus 5.5 at 40% less, OpenAI's harness behind an API, Okta's kill switch — and why the layer around the model is the one Europe has to own.
- [Anthropic Named Alibaba. Brussels Named the Risk.](https://www.onabout.ai/p/the-collapsed-distance): 151 million stolen exchanges, a one-person APT against European parties, agents that ignored their own operator — and von der Leyen's call to pace the frontier.
- [The Frontier Is Being Partitioned. Europe Just Bought a Seat.](https://www.onabout.ai/p/the-partitioned-frontier): Mistral's €3B round, GPT-6 Astra's capability gate, six Chinese labs named — and the CRA's 24-hour clock starts tomorrow.
- [Brussels Read the Post-Mortem.](https://www.onabout.ai/p/brussels-read-the-post-mortem): Last week this newsletter called AI Act enforcement "live but dormant." Two days later: information requests to 30+ AI companies, then ChatGPT pulled into the DSA's strictest tier, and the labs gating their own models before being asked. Supervision of frontier AI started this week — from both directions.
- [OpenAI's Agents Escaped. Hugging Face Caught Them](https://www.onabout.ai/p/openai-s-agents-escaped-hugging-face-caught-them): Back to the regular format, c'est la rentrée, August needs catching up on. Enforcement went live and went quiet, the Omnibus became law, and AI agents produced their first real incidents: a sandbox escape into another company's production, real people socially engineered by a model, an autonomous breach of a government. The catch-up, in one edition.
- [Governance You Can Deploy.](https://www.onabout.ai/p/governance-you-can-deploy): Build Lab 3 of 3. Policies live in documents; enforcement lives in infrastructure. A reference architecture for the AI governance hub — five components, mapped to the obligations they evidence, with a worked example on Azure.
- [The MCP Server Is Your New Perimeter.](https://www.onabout.ai/p/the-mcp-server-is-your-new-perimeter): Build Lab 2 of 3. Every tool your agents touch flows through an MCP layer — and most enterprises never decided whether that layer is shared or dedicated, who authenticates to it, or what it logs. The protocol won't decide for you.
- [Your Orchestration Pattern Is a Governance Decision](https://www.onabout.ai/p/your-orchestration-pattern-is-a-governance-decision): Build Lab 1 of 3. Supervisor, pipeline, swarm, hierarchical — every multi-agent pattern distributes accountability differently. Choose the pattern and you've chosen your audit story. Most teams choose by accident.
- [Sunday, Everything Changes. Here's Exactly What.](https://www.onabout.ai/p/sunday-everything-changes-here-s-exactly-what): On Sunday the Commission gains enforcement powers over general-purpose AI and Article 50 transparency applies across the EU. One essay: what applies, what doesn't, and what day one will actually look like.
- [An AI Broke Out of Its Sandbox. It Hacked a Real Company.](https://www.onabout.ai/p/an-ai-broke-out-of-its-sandbox-it-hacked-a-real-company): OpenAI's models escaped an isolated test environment, reached the open internet, and compromised Hugging Face's production infrastructure — to cheat on a benchmark. The lesson isn't about frontier labs. It's about where your own enterprise runs experimental agents.
- [Seven Builders. One Governor. Seventeen Days.](https://www.onabout.ai/p/seven-builders-one-governor-seventeen-days): A new study puts the EU's AI builder-to-governance hiring ratio at seven to one. Enforcement starts in 17 days. This edition is the minimum viable sprint for the time that's left.
- [Your Face Is Now the Price of Frontier AI](https://www.onabout.ai/p/your-face-is-now-the-price-of-frontier-ai): Anthropic can now ask consumer users for a government ID, a photo or video image, and facial geometry verification. Enterprise tiers are outside the policy — but your employees' personal accounts aren't. The identity layer of AI access just arrived, and it collides with GDPR head-on.
- [The AI Act Doesn't Define AI Agents.](https://www.onabout.ai/p/the-ai-act-doesn-t-define-ai-agents): Autonomous agents are the deployment story of 2026. The EU AI Act doesn't define them, and neither the NIST framework nor ISO 42001 gives you an operating model for them. That gap is where your accountability now lives.
- [August 2: The First AI Act Deadline That Touches Everyone](https://www.onabout.ai/p/august-2-the-first-ai-act-deadline-that-touches-everyone): High-risk obligations slipped to 2027. Transparency did not. The June 10 marking Code just told you exactly how to comply, and there's a December grace window.
- [Your Model Vendor Can Be Switched Off Without Notice](https://www.onabout.ai/p/your-model-vendor-can-be-switched-off-without-notice): Four days after launch, the US ordered Anthropic to cut off Fable 5 and Mythos 5, it immediately went dark for every customer on earth. Vendor outages have SLAs. This had none. Here's the diligence that makes your AI supply chain survivable.