If you use an Android phone and live in Europe, your phone's AI assistant has always been Google's. Not because you chose it. Because Google built Android, put Gemini at the center of it, and left no real system-level path for anything else.

That's changing. On July 16, the European Commission adopted binding orders under the Digital Markets Act (the EU's law governing how dominant tech platforms must behave) requiring Google to open 11 core Android features to rival AI assistants. The compliance deadline for most changes is August 1, 2027. By then, European Android users will be able to set Claude, ChatGPT, or another AI as their phone's actual assistant — not just an app you open separately, but a system-level AI that responds to your voice, reads your screen, and can act inside your other apps.

The US is not affected. This is EU law, applied to the roughly 700 million people who use Android phones across the European Economic Area. But it sets a precedent, and Brussels has a way of pulling the rest of the world's regulation in the same direction, slowly.

What "opening 11 features" actually means for you

Here's the object lesson, because the phrase "11 Android features" sounds technical and isn't.

Right now, when you say "Hey Google" on an Android phone, a few things happen that only Google's Gemini can do. It wakes up from your voice without you unlocking the phone first. It can see what's on your screen and respond to what's there. It can reach into your apps and do things: set an alarm, send a message, turn down the brightness. It's woven into the keyboard so it can suggest completions based on what you're typing.

A downloaded app like Claude or ChatGPT cannot do any of that from the system level. It lives in its own box. You open it, type into it, copy the answer out. Useful, but not what Gemini does when it's running as your phone's assistant.

The European Commission's order says Google has to give rival AI assistants the same access to those 11 features. Five categories: how you trigger the AI (voice wake words, home-button long press, CircleToSearch integration), what the AI can see (your screen, on-device app data, what you're typing), what the AI can do (control apps, change system settings like brightness or Do Not Disturb), background services, and the phone's own on-device computing power.

Starting in January 2027, there's a second piece: Google has to share anonymized query and ranking data with rival search engines and AI assistants. The practical effect: AI assistants that compete with Google will get access to more of the search signal that currently helps Google rank results. Over time, that makes competitor AI answers better.

Source spread

• European Commission — July 16 DMA Android AI interoperability order [official]. Primary source for the order, the 11-feature breakdown, and the January 2027 and August 2027 deadlines.
• EU DMA developer portal — Alphabet specification proceedings [builder]. The technical specifications that AI assistants must meet to qualify.
• TECHi — EU opens 11 Android features to rival AI assistants [skeptic]. Notes Google's track record with prior DMA compliance and the implementation risk.
• Nerd Level Tech — EU Android AI agent interoperability [builder]. Good on what the five feature categories actually are in practice.

What's real / what deserves a side-eye

What's real:

• The order is final. This is not a proposal. The European Commission adopted it on July 16 under Article 6(7) of the DMA, which is the article that lets the Commission issue binding interoperability requirements for gatekeepers. Google has to comply.
• The 11 features matter. Voice invocation and in-app control are the capabilities that separate a useful phone assistant from a chatbot you open and close. Getting those opened is substantive, not cosmetic.
• The January 2027 search data sharing is arguably the quieter but more consequential piece. More search signal means better factual answers from Claude, ChatGPT, and others.

What deserves a side-eye:

• Google's compliance history with DMA orders is mixed. The company has challenged, delayed, and technically-satisfied prior obligations in ways that required follow-up enforcement. Expect the letter to be followed; expect the spirit to take longer.
• "Certified AI assistants" is doing a lot of work. The order applies to compliant third-party assistants, meaning there will be a certification or qualification process. How hard that process is determines who actually benefits from this.
• If you're in the US, none of this applies yet. American Android phones are unaffected. Google has shown no sign of voluntarily extending these changes beyond the EEA.

What to do about it

A short list. Most of these are "wait and see," which I know is unsatisfying.

• If you're in Europe on Android: nothing changes before August 2027. This isn't live yet. The order exists, the deadline is set, but Google hasn't shipped the compliance.
• Watch what "certified AI" ends up meaning. When Google publishes its implementation specs for how rival AIs can apply to access the 11 features, that's the real test of whether this order has teeth.
• Try Claude or ChatGPT on your phone now, as a standalone app. The system-level integration you'll get in 2027 will be better, but you can already get a sense of whether you'd want it as your primary assistant. If you don't prefer it over Gemini without the deep access, you probably won't after either.
• If you're in the US: bookmark this. The EU regulatory framework tends to pull other jurisdictions toward it, slowly. Not guaranteed, but worth watching.

Further reading

• European Commission — July 16 DMA Android AI interoperability order
• EU DMA developer portal — Alphabet specification proceedings for AI interoperability
• TECHi — EU opens 11 Android features to rival AI assistants
• Nerd Level Tech — EU Android AI agent interoperability DMA
• Model Diplomat — EU forces Google to open Android AI to third parties