What happens when the most capable AI assistant you can get also needs your calendar, your email, your habits, and your location history to function? That's the deal Google formalized at I/O 2026.
The two headline features from the event are Gemini Spark and Daily Brief. Gemini Spark is an always-on AI agent that helps organize upcoming events, surfaces commitments you may have overlooked, and anticipates what you need next. Daily Brief delivers a personalized morning rundown of news, tasks, and updates relevant specifically to you. According to The Verge's coverage of the event, both features are built on the premise that users will extend Google the trust it needs to access that data.
The Actual Trade-Off
This isn't abstract. For Gemini Spark to remind you to prep for a meeting, it needs calendar access. For Daily Brief to summarize what matters to you, it needs to know who you are and what you follow. Google already holds most of this data for users of Gmail, Google Calendar, Maps, and Android. What's changing is that the company is now explicitly tying its most capable AI features to how much of that data you authorize it to analyze.
For people already operating inside Google's ecosystem, the incremental ask is small - the data exists, you're just granting permission for AI to connect the dots across it. For users who've deliberately kept Google as a search engine while using Apple Mail or a separate calendar, these features create a genuine decision point. Accessing the full capability means broadening your data footprint with Google.
Who Google Is Betting Against
Apple is running a different strategy. Apple Intelligence processes as much as possible on-device, meaning your data stays on your hardware rather than flowing to Google's servers. Microsoft's Copilot operates on enterprise data under different contractual privacy terms. Google is betting that its server-side AI is capable enough that users will accept cloud processing in exchange for the capability.
There's a reasonable case Google wins that bet. Its models are genuinely strong, its infrastructure is mature, and a large share of the world already uses Google as a daily operating layer over their professional and personal lives. For those users, Gemini Spark and Daily Brief are a natural extension of existing behavior.
The harder question is whether Google has earned the trust it's now explicitly asking for. The company has a mixed record: technically strong security, but a long history of surprising users with how data gets used - killed products, ad targeting, policy updates that reset reasonable expectations. "We won't use your AI-processed data to train models without consent" is the kind of assurance that sounds clear until the terms of service get updated three years from now.
Gemini Spark and Daily Brief aren't previews. Google described them in implementation detail at I/O, which means users will be making these opt-in choices soon. How many actually grant the access will tell us more about Google's trust deficit than any survey could.