MOUNTAIN VIEW, California — January 15, 2026, 08:30 PST
- Google introduced a beta “Personal Intelligence” feature for the Gemini app, tapping into data from Gmail, Photos, Search, and YouTube history.
- According to the company, access is now rolling out in the U.S. for eligible Google AI Pro and AI Ultra subscribers, though the feature remains off by default.
- This update advances Gemini by delivering more personalized and proactive responses, while also emphasizing stronger privacy controls and building user trust.
On Wednesday, Google started deploying “Personal Intelligence,” a beta capability within its Gemini AI that connects to a user’s Gmail, Google Photos, Search, and YouTube history to customize responses, the company announced. The feature is disabled by default and will be available over the next week to qualifying Google AI Pro and AI Ultra subscribers in the U.S., with plans for a wider rollout later. Blog
Why this matters now: the competition has moved beyond just having the best model to how seamlessly it can integrate into someone’s everyday routine without causing alarm. “Personal” is the key selling point — but also the biggest vulnerability.
Google aims to evolve Gemini from a simple question-answering chatbot into a platform that links together the information users already have stored within Google. This approach directly challenges competitors like OpenAI and Anthropic, which excel in raw model responses but lack access to the same breadth of consumer data.
Simply put, Google’s Gemini can draw information simultaneously from multiple sources—like an email, a photo, or a watch history entry—to respond to a single prompt. Josh Woodward, the VP in charge of the Gemini app, highlighted the feature’s “two core strengths” as “reasoning across complex sources” and extracting “specific details” from personal content. According to TechCrunch, Gemini can even generate proactive links from that data without needing explicit directions on where to search.
Google demonstrated the “tire shop” example, where Gemini spots a vehicle detail and then adds personal context, such as trip photos, to refine the options. According to 9to5Google, Gemini will also highlight where it pulled critical info from connected sources, helping users verify the answers more easily.
The privacy details get complicated here. According to a Google help page, data from connected apps can be used to personalize Gemini and “improve Google services,” which includes training generative AI models—depending on your settings. It also mentions that some of this processed data might be checked by human reviewers. While Gemini doesn’t train models “directly” on a full Gmail inbox or entire Photos library, it can use summaries, excerpts, and inferences generated to respond to prompts. Google Help
Still, the system isn’t perfect—Google warns users against putting too much faith in it. Woodward told The Verge that “over-personalization” can cause the model to link unrelated topics, and said timing and nuance remain challenging, especially with changing relationships. The Verge also noted that the feature runs on Google’s Gemini 3 models and requires users to opt in, selecting which apps they want to connect. The Verge
Google is pushing Gemini further into consumer tech by expanding its partnerships. According to Reuters, Apple plans to integrate Google’s Gemini models into a redesigned Siri set to launch in 2026, thanks to a multi-year agreement. This move could significantly increase Gemini’s presence across Apple devices. Analyst Parth Talsania from Equisights Research sees it as a shift that positions OpenAI more in a supporting role. Meanwhile, Tesla CEO Elon Musk called the deal an “unreasonable concentration of power.” Reuters
Google is weaving Gemini into subtle product updates that tweak daily workflows without fanfare. TechCrunch revealed the company is upgrading the Google Trends Explore page on desktop, adding Gemini-powered tools that automatically spot and compare related trends.
The next phase hinges less on demos and more on user habits: will people actually enable the feature, and will the “personal” layer come across as helpful or invasive? Trust plays a big role too — one damaging privacy scandal or a handful of notable errors based on personal data could quickly stall adoption.
