Gemini 2.5 Pro Hits Google Search

Mind the bots, Tech Insiders.

AI assistants, AI hunters, AI layoffs—machine brains are everywhere except the unemployment line.

Let's sift the perks from the pitfalls.

Here's what you need to know today:

Gemini 2.5 Pro Supercharges Google Search

Hate phone calls? Google's bot now dials so you don't have to.

Google is rolling out Gemini 2.5 Pro inside AI Mode, letting paid AI Pro and AI Ultra subscribers tap its most powerful model for tougher math, coding, and research questions.

A new Deep Search option—also powered by 2.5 Pro—fires off hundreds of behind-the-scenes queries, then assembles a fully cited report in minutes, turning tedious desk research into a one-click brief.

The splashiest trick? Agentic, AI-powered business calling.

Search for something like "pet groomers near me," hit "Have AI check pricing," answer a few prompts, and Google rings around to collect quotes and availability, clearly identifying itself as a bot. The feature began rolling out nationwide on Wednesday to all US users, while subscribers get higher usage caps.

Google frames these perks as early-access previews; once the kinks and ethics debates settle, expect broader availability. Businesses can opt out in their Business Profile settings—but if they don't, they should brace for more robo-inquiries.

Why it matters: Less hold music, more answers—AI could finally lighten your chore list. Yet, every automated call provides Google with fresh pricing data and greater influence over local commerce. Convenience comes at the cost of ceding a little more ground to Big Search.

Share this newsletter!

Pulse Check

Would you let Google's AI handle your next service call?

Results from Yesterday's Pulse Check

How do you feel about Google's new Discover summaries?

Insider Intel

AI Agent Preempts Real-World SQLite Exploit

When your bug hunter never sleeps—because it's literally named Big Sleep.

Google's Big Sleep agent bagged a new trophy: CVE-2025-6965, a high-severity SQLite memory-corruption flaw (CVSS 7.2) that threat actors already had in their sights. Fueled by a mix of threat-intel breadcrumbs and self-directed code spelunking, the agent predicted exploitation was imminent and alerted engineers before any in-the-wild hits occurred.

The episode reinforces a trend: AI can shift security from a reactive patch-and-pray approach to a proactive seek-and-disarm strategy. Google is donating Secure AI Framework data to the Coalition for Secure AI, allowing others to replicate the win—and it's showcasing new agentic forensics tools such as Timesketch (now Sec-Gemini-powered) to slash incident-response time.

Skeptics still want more transparency on how Big Sleep reasons and on guardrails against rogue actions. But the takeaway is clear: automated vulnerability discovery just shortened attackers' leashes.

Sweet dreams are made of patched bugs—no lullaby required.

Flash Bytes

Security Alerts

Chrome Patches Sandbox-Escape CVE-2025-6558

For the fifth time this year, Google rushed out a fix for an in-the-wild zero-day—this one in ANGLE/GPU code that lets a malicious page jump Chrome's sandbox. Discovered by Google TAG on June 23, the flaw (high-severity CVSS 8.8) is already weaponized, likely by state-sponsored actors.

Update to Chrome 138.0.7204.157 / .158 on Windows and macOS, or .157 on Linux and Android as soon as possible, and expect Edge, Brave, and other Chromium browsers to follow.

Pro tip: If the "About Chrome" tab feels like a weekly chore, enable auto-update and caffeinate—zero-days wait for no one.

Scattered Spider's 'Frequent-Flier' Phase Gets Uglier

We flagged the crew's airline antics back on July 7—Microsoft's new report just fills in the grisly details. 🕷️✈️

The social-engineering specialists known as Scattered Spider (aka Octo Tempest), already tied to hacks at Hawaiian Airlines, WestJet, and a Qantas call-center vendor, have expanded their flight path. Redmond's analysts say the gang now:

That "ground-first, cloud-later" pivot flips the usual playbook, buying attackers more time inside fleets of virtual machines.

Why it matters: Boarding groups are stressful enough without arachnids in the jet bridge. Patch identity workflows, segment ESX clusters, and remind support staff that spiders don't need extra legroom.

Industry Shakeups

Scale AI Trims 14% After Meta Buy-In

Just weeks after Meta bought a 49% stake and hired founder Alexandr Wang, data-labeling heavyweight Scale AI is laying off 200 full-time employees and ending work with 500 contractors.

Interim CEO Jason Droege blamed overzealous hiring and shifting demand as marquee customers such as Google and OpenAI pulled back, uneasy about Meta's new line of sight.

Scale says it will refocus on its most profitable niches—including code, language, and audio datasets—and beef up government and enterprise sales later this year.

With rivals circling and clients skittish, Scale's next trick is proving it can thrive without the charismatic founder who just joined its biggest investor.

Lesson: When your CEO jumps to a rival, tidy the org chart before the chairs start spinning.

Curious about where AI is really headed?

The Neuron cuts through the noise to bring you smart, hype-free takes on the latest AI trends, tools, and breakthroughs. Join 500,000+ professionals from top companies like Microsoft, Apple, Salesforce and more.

Advertise in Daily Tech Insider!

Daily Tech Insider is a TechnologyAdvice business

© 2025 TechnologyAdvice, LLC. All rights reserved.

TechnologyAdvice, 3343 Perimeter Hill Dr., Suite 100, Nashville, TN 37211, USA.

Want to change how you receive these emails? Update your preferences or unsubscribe here.