YouTube leans into genAI with new tools for creators - DTNS 5105

Meta leaked its monocular smartglasses ahead of Meta Connect 2025, and Dr. Niki talks all about Robot Antelopes.

Starring Jason Howell, Tom Merritt, and Dr. Niki.

JASON: This is the Daily Tech News for Tuesday, September 16, 2025. We tell you what you need to know, follow up on the context of those stories and help each other understand.

TOM: Today Dr Niki tells us about Robot Antelopes and YouTube increases its appeal to live streamers and podcasters and even moves ads out of the main video window.

I’m Jason Howell,

I’m Tom Merritt.

JASON: Let’s start with what you need to know with the big story.

BIG STORY

YouTube Live gets a major update, introduces dual-format streaming, minigames, and more
YouTube announces new generative AI tools for Shorts creators
YouTube to use AI to help podcasters promote themselves with clips and Shorts
YouTube just announced a bunch of AI tools for creators

The Made on YouTube live event took place this morning, and a slew of new features, many AI-driven, are coming to the platform. And there's a lot, so buckle up.

YouTube Live is rolling out its biggest update yet.

• Free mini-games called Playables that can be played during live streams. YouTube sees this as an entry point for new streamers to build a fanbase.

• Support for dual-format streaming, so creators can simulcast in horizontal and vertical video with a single unified chat.

• A new AI-powered highlights tool that turns key livestream moments into shareable shorts.

• New side-by-side ad formats that can monetize streams without interrupting the live content.

• Channels with memberships can switch from public to members-only streams on the fly.

As for Shorts, it’s all about generative AI tools.

• A custom version of Veo 3 Fast that creates at 480p with sound.

• Apply motion from a source video to an image, so you can make the person in the image dance, for example.

• Apply sound or dialogue from one video into a new Short as the soundtrack source, powered by Google’s Lyria 2.

• Edit with AI feature places raw footage into first drafts with music, transitions, and voiceover options.

As for podcasts:

• Turn episodes into video clips and YouTube Shorts, including audio-only shows

• Video podcasts will get AI-assisted clip suggestions

• Audio podcasts can generate customizable videos with the tools.

TOM: DTNS is made possible by you the listener. Thanks to
Michelle Sirjue
Kirk Steffensen
MirandaJanell
New Patrons: TM, Joe, Kenton, and Kirsty
And a raise from GjamesB

JASON: There’s more we need to know today, let’s get to the briefs.

BRIEFS

OpenAI is building a ChatGPT for teens
Building towards age prediction
Teen safety, freedom, and privacy
TOM: Ahead of testifying to the US Senate Judiciary Committee, OpenAI announced that it will soon launch new user policies aimed specifically at those between 13 and 18 and meant to address mental health and legal concerns tied to the use of chatbots by minors. By the end of the year, OpenAI will have built "an age-prediction system to estimate age based on how people use ChatGPT." It will default to teen protections when age can't be determined and "give adults ways to prove their age to unlock adult capabilities." Enhanced parental controls are also coming soon. When a teen links their account to a parent's, the parent can control chat history storage, set blackout hours, and get alerts when distress is detected. OpenAI CEO Sam Altman wrote, "We prioritize safety ahead of privacy and freedom for teens; this is a new and powerful technology, and we believe minors need significant protection."

What to expect from Meta Connect 2025 this week: Ray-Ban smart glasses, Hypernova, more
'Meta Ray-Ban Display' Glasses Design & HUD Clips Leak Ahead Of Connect
JASON: Meta Connect 2025 begins tomorrow, September 17, and this guy named Jason Howell wrote for ZDNet about what to expect at the event! OK, no more third person.

So I wrote about rumors, and then Meta leaked a video on its YouTube channel confirming a bunch of them! The now-removed video showed off an assortment of glasses, including the Ray-Ban and Oakley HSTNs we already have. Beyond that, Oakley Meta Sphaera frames with added eye protection and a centered camera above the nose are perfect for sports scenarios. But the big leak is Meta's monocular frames in a partnership with EssilorLuxottica, black frames that feature a fixed HUD that can display Meta AI content, navigation maps, and more. Also teased was the sEMG wristband that's meant for interactivity and control of the HUD glasses.

Beyond that, no new Quest hardware is expected, though ASUS may show off its ROG Tarius headset, a premium alternative to Meta's VR hardware. The event begins tomorrow at 5 pm PT with a keynote by Mark Zuckerberg.

Internet Archive’s big battle with music publishers ends in settlement
TOM: The Internet Archive, along with several major music publishers, has settled a lawsuit over the Great 78 project, which had plans to digitize 3 million old shellac records containing rare recordings. Details of the settlement were not shared, but record labels had ultimately claimed up to $700 million in damages, which would have ruined the company had it succeeded. The Internet Archive lost a similar case with book publishers last year, which also resulted in an undisclosed settlement.

Apple backports zero-day patches to older iPhones and iPads
JASON: Apple issued security updates for older iPhones and iPads to patch a zero-day vulnerability that had already been fixed on newer devices. The flaw is related to the Image I/O framework and can allow attackers to exploit memory corruption with malicious image files, which can lead to remote code execution. iOS and iPadOS versions 15.8.5 and 16.7.12 are now addressed with the security update.

"Apple Preparing Four New 2nm Chipsets In 2026, With At Least Two Of Them Adopting An Advanced Packaging Technology"
TOM: TSMC will begin producing chips on its 2nm process in Q4 this year, and China Times' sources say Apple has ordered use of about half of the company's new 2nm capacity. Reportedly, that would be for A20 and A20 Pro chipsets for use in next year's iPhone, but also possibly for a Mac processor and a Vision Pro co-processor, both to come in 2026. Apple is also reportedly adopting TSMC's Wafer-Level Multi-Chip module packaging or WMCM. Packaging combines multiple parts together, so the CPU, GPU, and DRAM can be integrated into a smaller package, which makes it more power efficient and in some cases, easier to cool. Chips made on the 2nm process also benefit from power efficiency as well as higher performance. So you can expect any device, from Apple or anyone else, made on this process to have longer battery life, be more powerful, and easier to make lighter and thinner. It'll also cost more. TSMC's 2nm process is estimated to cost around $30,000 per wafer, with yields usually fewer than 1,000 chips per wafer.

This is the first Roku TV smart projector
JASON: Aurzen launched the first projector running Roku's TV OS, and it's called the D1R Cube. It can cast images up to 150 inches but has limited brightness at 330 ANSI lumens, necessitating a very dark room and reflective screen to get the best output. It can take in 4K content but only outputs to 1080p resolution. The projector has an automatic focus feature along with keystone adjustments, and also includes a pair of 5-watt speakers. The D1R Cube is now available for $169.99.

TOM: Those are the essentials for today. Let’s dive a little deeper.

DIVE A LITTLE DEEPER

JASON: Want to collect data on Tibetan antelopes? Why not become one? Or better yet make a robot that they can accept.

TRT - 6:25

PROMO

TOM: Join in the conversation in our Discord which you can join by linking to a Patreon account at patreon.com/dtns

HELPING EACH OTHER UNDERSTAND

JASON: We end every episode of DTNS with some shared wisdom. Monday, Tom and Robb were talking about the perception that might come with a change in TikTok US ownership. AnonJr has some further thoughts along those lines.

TOM: AnonJr writes:
If I remember correctly, someone did an experiment where they took Google search results and dressed it in a Bing results page, and vice versa, and there was a strong preference for the "Google" results. I wonder if something similar could be done with TikTok and Instagram. Though the biggest difference is likely in the culture of what is created, shared, and consumed. E.g. would Instagram's algorithm produce TicTok like results if the input and consumption patterns were the same. Random 1st cup of coffee thought.

JASON: What are you thinking about? Got some insight into a story? Share it with us feedback@dailytechnewsshow.com

JASON: Thanks to AnonJr and Dr. Niki for contributing to today’s show. And thank YOU for being along for Daily Tech News Show. You can keep us in business by becoming a patron, atPatreon.com/dtns