Good morning, today's Factwire has Grok's top-ranked new voice, a Meta model that runs on two Macs, and a scam check for texts.
Grok's new voice tops an outside ranking at the same price
In one line: xAI released Grok Voice 3.0, the voice behind Grok Bot's voice mode and Grok in Tesla cars.
What happened:
Artificial Analysis, an independent tester, ranked it first with a score of 86.9, ahead of voice models from OpenAI and Google.
xAI says it answers in under a second, and app makers pay the same 8 cents a minute as before.
Starlink call centers use it for 20,000 calls a day, and xAI says about 70% get solved with no person stepping in.
Why it matters: If you talk to Grok Bot out loud, this is the voice answering you.
xAI, Oct 14, 2026.
Factwire's take: Voice is where AI starts to feel like a helper instead of a search box. A first-place score from an outside tester means more than a company grading its own work. Still, the best test is yours, so ask it something messy and see if it keeps up.
Hit reply: do you talk to your AI out loud, or do you still type?
Meta's new open AI model can run at home on two Mac Studios
In one line: Meta released Muse Spark Open, which it calls the most capable US open-weight model so far, and posted its files for download.
What happened:
Meta says it is the first of its kind that can run on a desk, using two Apple M5 Ultra Mac Studios.
It is built for coding and long, multi-step jobs, and services like OpenRouter also run it online.
Why it matters: Few people will run this at home yet, but it shows strong AI can now work on a desk, not only in a data center.
Meta AI Research, Oct 15, 2026.
Not sure about a text? Your Android phone can check it for scams
In one line: Google says Circle to Search on Android can look over a suspicious text and point out signs it may be a scam.
What happened:
You press and hold the home button or navigation bar, circle the message, and get an AI Overview with tips and suggested next steps.
If it shows red flags, Google suggests deleting the message and blocking the number, and says results and availability may vary.
Why it matters: Scam texts keep getting harder to spot, and this gives you a second opinion before you tap a link or share anything personal.
Google, Oct 15, 2026.
Start here
Try it this week: check a text with Circle to Search
Open the text you're unsure about on your Android phone.
Long-press the home button or the navigation bar at the bottom of the screen.
Draw a circle around the message and read the AI Overview that appears.
If it shows red flags, don't tap any links, then delete the message and block the number.
Shortcut: Practice once on a text you know is safe, so the steps feel easy when a real scam shows up.
Wires
OpenAI says its GPT-Live-1.1 update gives ChatGPT Voice new voices, better accents and fewer "checking..." pauses, rolling out to all users. OpenAI on X, Oct 15, 2026.
Amazon's $74.99 Bee Frontier wearable turns your conversations into summaries and to-dos, and Amazon says the audio is processed, then deleted. Amazon, Oct 15, 2026.
Anthropic gave six staff from METR, a nonprofit safety tester, employee-like access to check its AI, and Anthropic can't veto what METR publishes. Anthropic, Oct 15, 2026.
Replit CEO Amjad Masad says OpenAI is now the preferred partner powering Replit's coding Agent, and users can still pick Anthropic models and others. Amjad Masad on X, Oct 17, 2026.
A machinist posting as @SafetySquintz says Grok Bot helped overhaul the computer network at their shop, including automatic off-site backups and new security systems. @SafetySquintz on X, Oct 15, 2026.
A Harvard Shorenstein Center study found 29% of women have never used AI, compared with 22% of men. Shorenstein Center at Harvard, Oct 14, 2026.
Using AI for something? Hit reply and it might land in Reader's corner.
Hit reply, what should we cover next?
Know someone who keeps asking you about AI? Forward them this email. They can join Factwire free at www.factwireai.com/subscribe
Next drop on Friday, enjoy your week.