Gemini passes a billion, Zoom patches: August 12 brief

Gemini now reaches more than a billion people each month, so being understood by AI is becoming a customer-acquisition issue for almost every business. Spotify is also drawing a harder line around synthetic identities. First, though, teams that use Zoom have a simple security job today: update the app before the next call.

Google takes Gemini past a billion monthly users

Google said on August 11 that the Gemini app has passed 1 billion monthly users, making it the fastest-growing product in the company's history. A monthly user is someone who opens the app at least once during a month. Google also said Gemini produces more than 150 million images each day, and that over 100 million active users access it on Apple phones and tablets.

This scale changes where buying research begins. A customer can now ask Gemini which local company to call, which software fits a small team, or what product solves a specific problem. The assistant may answer before that person visits a normal search-results page. Our guide to how AI recommends a business explains the sources these systems tend to use.

The practical move is to check what Gemini says about your company, your main service, and two close competitors. Look for stale hours, missing products, weak reviews, or a description you wouldn't use yourself. Then correct the underlying business listings and website pages. This is an information-cleanup job, not a prompt-writing contest.

Google's August 11 usage update is a platform claim, so the figures haven't been independently audited. The direction is still clear. AI assistants have enough reach to deserve a place beside ordinary search in a monthly visibility check. Businesses that sell products can start with our plain-English look at ChatGPT shopping discovery, since the same clean-data principle applies.

Spotify removes AI personas from recommendations

Spotify said on August 11 that profiles presenting a computer-made person as an artist will receive an AI Persona badge starting in mid-September. The badge will appear on profiles, in search, and beside tracks in playlists. Spotify will also keep those profiles out of its editorial and automated recommendations by default, unless a listener has chosen to follow them.

An automated recommendation is a song or account selected by software based on a listener's behavior. Losing that distribution can make a large audience much harder to reach. Spotify will accept self-disclosure, run its own reviews, and offer an appeal when its team applies the label. The rule concerns the public identity behind an artist, rather than every use of AI in making music.

The wider signal matters beyond music. Platforms are starting to separate AI-assisted creative work from invented people, then attach visibility consequences to the second category. A brand using a synthetic spokesperson should disclose that choice clearly and keep proof that it owns the images, voice, and name. Our AI brand voice guide covers the safer side of the line: using AI to help express a real company's established identity.

Spotify's official announcement gives affected artists a month to review their profiles. We think brands should read the policy as an early warning. A fictional face that looks human may save a shoot today, then lose reach when the next platform copies this approach.

Zoom closes a silent device-takeover flaw

Zoom published three security bulletins on August 11 for flaws in its annotation feature, the tool that lets meeting participants draw or type over a shared screen. The most serious weakness could let someone in a meeting run harmful code on another participant's device without that person clicking or downloading anything.

Researchers at A Security reported that AI helped them find the flaw and build a working attack with fewer than 20 prompts in under 24 hours. Zoom has issued client and server fixes. Its security bulletin page lists the affected issues as CVE-2026-53413, CVE-2026-53414, and CVE-2026-53415. A CVE is the public tracking number assigned to a software security problem.

Update Zoom Workplace on every company device before another meeting. Organizations that centrally manage computers should confirm the installed version rather than trusting staff to notice an update prompt. The research team says the protected branches are version 7.1.5 or 7.0.6 and later. This one is urgent because a normal-looking meeting invitation can be enough to create exposure.

Worth watching

• OpenAI released a preview of its ChatGPT desktop app for Linux, the operating system common on technical teams' computers.
• Apple is testing photo metadata that could help people check whether an image came from an iPhone camera, a useful trust signal for newsrooms and brands.
• Spotify plans to let listeners report suspected AI Persona profiles in the coming months, which will test how well its appeals process handles mistakes.