Google Warns EU Android AI Interoperability Rules Could Weaken Mobile Security

Google is pushing back on the European Commission's new Android interoperability requirements for AI services, arguing that the implementation phase needs stronger security guardrails before third-party agents receive deeper access to phone capabilities. In an August 5 post from Android security leaders Dave Kleidermacher and Eugene Liderman, the company said the EU Digital Markets Act process has moved from policy into technical execution, where details such as approval, suspension, and revocation of agent access will decide whether the rules increase choice without opening new abuse paths.

The dispute centers on specification proceedings the Commission opened for Android features relevant to AI services. The Commission says its final decision, adopted on July 16, sets measures Google must implement to enable effective interoperability. Google says it supports Android's open ecosystem, but objects to mandates that could let user-downloaded AI agents reach sensitive device capabilities more deeply than ordinary apps do today.

Why the Security Debate Matters

Google's main concern is that AI agents are not conventional mobile apps. A normal app is usually constrained by Android's sandbox and permission model. An agent with broader system hooks can observe context, automate actions, and act across services, which makes the consequences of compromise or manipulation more serious. Google specifically flagged access to ambient data such as always-on microphone input, camera data, and on-screen content, as well as high-risk capabilities like screen automation.

The company also tied the issue to newer agentic-AI risks, including indirect prompt injection, where hidden instructions inside external content can steer an autonomous system away from the user's intent. Google's argument is not that rival AI services should be excluded from Android, but that access should be governed by enforceable qualification processes and platform-level controls. It says device makers and Google need authority to maintain the certification lifecycle rather than relying only on third-party certification bodies.

Competition Meets Platform Control

The Commission's position is rooted in the DMA's goal of preventing gatekeepers from reserving important platform capabilities for their own services. In practice, the Android decision could shape how assistants from companies such as OpenAI, Anthropic, Perplexity, and others integrate into European phones. If the implementation is broad, rival agents may be able to invoke more device functions and compete more directly with built-in assistants.

For developers and users, the story is less about one company complaint than about a policy tradeoff now becoming engineering work. Europe wants more interoperability and less self-preferencing by dominant platforms. Google is warning that mobile AI agents raise a different threat model because they combine sensitive context, automation, and probabilistic behavior. The strongest outcome would give users more meaningful assistant choice while preserving clear limits around microphones, screens, account data, and automated actions.

The next phase will be watched closely because Android is both a major consumer platform and a test case for how competition law handles AI agents. The rules that emerge in Europe may influence how mobile operating systems expose privileged capabilities elsewhere, especially as AI assistants become more capable and more deeply embedded in everyday device workflows.

Sources

• blog.google
• digital-markets-act.ec.europa.eu

Cover photo by Dan Nelson on Pexels, used under the Pexels License.