Tech Brew Ride Home - Thu. 12/07 – Zuck Keeps A Promise

Episode Date: December 7, 2023

Meta has kept a longstanding promise to encrypt Messenger by default. Apple is coming clean about the notifications spying thing. AMD’s answer to Nvidia. We know when the next major version of Windows is coming and, surprise, it’s got AI all over it. And Apple releases its first little AI hint. Sponsor: Dragonball Legends Links: Messenger is finally getting end-to-end encryption by default (The Verge) Federal government is using data from push notifications to track contacts (Washington Post) AMD unveils Instinct MI300X GPU and MI300A APU, claims up to 1.6X lead over Nvidia’s competing GPUs (Tom's Hardware) Meta launches a standalone AI-powered image generator (TechCrunch) PlayStation keeps reminding us why digital ownership sucks (The Verge) EXCLUSIVE: Microsoft readies 'groundbreaking' AI-focused Windows release as new leadership takes the helm (Windows Central) Apple launches MLX machine-learning framework for Apple Silicon (ComputerWorld) Is Elon's AI Copying Me? Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the TechMeme right home for Thursday, December 7th, 2023. I'm Brian McCullough today. Meta has kept a longstanding promise to encrypt Messenger by default. Apple is coming clean about the whole notifications spying thing, AMD's answer to Nvidia. We now know when the next major version of Windows is coming and surprise it's got AI all over it, and Apple releases its first little AI hint. Here's what you miss today in the world of tech. Meta has rolled out end-to-end encryption to Messenger by default. for one-on-one chats and calls, thereby fulfilling a long-term promise. End-to-end encryption for group chats remains opt-in, but hey, they finally shipped, quoting

The Verge. Encrypted chats were first introduced as an opt-in feature in Messenger in 2016, but after a long wind-up, end-to-end encrypted messages and calls for conversations between two people will now be the standard going forward. This has taken years to deliver because we've taken our time to get this right. Laura Donna Cresson, VP of Messenger, said in a statement shared with The Verge. Our engineers, cryptographers, designers, policy experts, and product managers have worked tirelessly to rebuild Messenger features from the ground up, end quote.

According to Cresson, you won't sacrifice Messenger features when using encrypted chats, so you'll still be able to use things like themes and custom reactions. However, Crizzan notes that it may take some time for all Messenger chats to switch over to default encryption. While this is a good step, end-to-end encryption for group messenger chats is still opt-in for now. Instagram messages are also not encrypted by default, though Meta said in August that would happen shortly after the rollout of default private messenger chats. CEO Mark Zuckerberg announced in 2019 that the company planned to move toward encrypted ephemeral messages across its messaging apps. I believe the future of communication will increasingly shift to private

encrypted services where people can be confident that what they say to each other stay secure and their messages and content won't stick around forever. He wrote in a Facebook post, This is the Future. I hope we will help bring about, end quote. Follow up to a segment from yesterday. Apple guidelines now say that U.S. law enforcement can obtain the Apple ID associated with a push notification token via a subpoena. Google apparently requires a court order as well, quoting the Washington Post. Government investigators in the United States have used push notification data to pursue people of interest, Senator Ron Wyden said in a letter Wednesday to the Justice Department revealing for the first time a way in which Americans can be tracked through a basic service provided

by their smartphones. Apple's law enforcement guidelines, the company's rules for how police and government investigators should seek user information. Now note that a person's Apple ID associated with a push notification token can be, quote, obtained with a subpoena or greater legal process, end quote. Neither Wyden nor Apple detailed how many notifications had been reviewed, who had been targeted, what crimes were being investigated, or which governments had made the requests. Google said in a statement that it publishes transparency reports sharing the number and types of government requests for user data it receives, and that it shares Wyden's, quote, commitment to keeping users informed about these requests. The companies, Wyden wrote, told members of his staff that

any, quote, information about this practice was restricted from public release by the government. Wyden pushed the Justice Department to repeal any policies forbidding the companies from discussing the surveillance practice. Apple and Google should be permitted to be transparent about the legal demands they receive, particularly from foreign governments, just as the companies regularly notify users about other types of government demands for data, he wrote, end quote. We've been talking about how everybody wants AI chips. InVIDIA is making them as fast as they can. Other companies are attempting to solve their own bottlenecks by designing their own chips. But what about other traditional chip players? Well, AMD has launched the Instinct M-I-Cube.

300x and MI300A AI accelerators and claims the MI300X delivers up to 1.6x more performance than NVIDIA's H-100 HGX in inference workloads. Quoting Tom's hardware. AMD launched its Instinct MI300X AI accelerator and the instinct MI300A, the world's first data center APU, during its advanced AI event here in San Jose, California, as it looks to capitalize on the booming generative AI and HPC market. AMD forged its MI-300 lineup using the most advanced production technologies ever pressed into mass production, employing new techniques like its 3.5D packaging to produce two multi-chip behemus that it says provide Nvidia beating performance in a wide range of AI workloads. AMD isn't sharing pricing for the new exotic silicon, but the products are now

shipping to a wide range of OEM partners. The instinct MI-300 is a game-changing design. The data Center APU blends a total of 13 chip, many of them 3D stacked to create a chip with 24 Zen 4 CPU cores, views with a CDNA3 graphics engine, and eight stacks of HBM3. Overall, the chip weighs in with 153 billion transistors making it the largest chip AMD has ever made. AMD claims this chip delivers up to 4x more performance than Nvidia's H-100 GPUs in some workloads and touts that it has twice the performance per watt. AMD says its instinct MI300X GPU delivers up to 1.6x more performance than the

NVIDIA H100 in AI inference workloads and offers similar performance in training work, thus providing the industry with a sorely needed high performance alternative to NVIDIA's GPUs. Additionally, these accelerators come with more than twice the HBM3 memory capacity than NVIDIA's GPUs, an incredible 192 gigabytes apiece, enabling its MI300X platforms to support, more than twice the number of LLMs per system, as well as running larger models than NVIDIA's H-100-HGX, end quote. Microsoft, Meta, Oracle, and OpenAI all say they plan to use the MI-300X. Microsoft plans to offer access to the chips through Azure. I feel like there are announcements like this every day recently, so real quick, in response to Google's

Gemini launch, Meta has introduced a new standalone generative AI experience on the internet called Imagine with Meta. This innovative tool enables users to generate images by describing them using natural language, much like OpenAI's Dali, Mid Journey, and Stable Diffusion. Imagine with meta leverages meta's emu image generation model to produce high-resolution images from textual prompts. Currently, it's available for free to users in the United States and generates four images per prompt. Meta has expanded the use of Imagine outside of chat interactions and is now accessible on the web for creating free images. In the near future, meta plans to add watermarks to the content generated by Imagine

for increased transparency and traceability. These watermarks, although initially invisible, will be generated using AI models and detectable with corresponding models, though it's unclear whether the detection model will be made public. The tech industry is facing growing pressure to clearly indicate when AI-generated content is involved, especially in light of concerns related to deepfakes and AI-generated inappropriate imagery. I've done a million stories like this over the years, but it's always good to remind you that in our new era of digital media, you don't own anything. Quoting the Verge. Last week, Sony said that because of content licensing arrangements, PlayStation users

wouldn't be able to watch Discovery content they've purchased, and that the content would be removed from their libraries as of December 31, 2023. The resulting list of shows that will suddenly disappear because of corporate agreements is very long. Shows disappearing from streaming services is commonplace, but in this case, people are losing access to shows they bought to watch on demand whenever they wanted. Then on Monday, many users were unexpectedly banned from their PlayStation network accounts, meaning that not only were they blocked from playing multiplayer games or using cloud streaming, but they were also locked out of games they purchased digitally from Sony's PlayStation Marketplace. Affected users who may have spent years building a robust

digital library were suddenly left without access to content they had bought through no fault of their own. It appears that Sony has since restored account access to people who were accidentally banned, but the company hasn't explained what happened or said how it might prevent similar unexpected bans in the future. The ephemorality of digital ownership isn't a new issue, even though downloading and accessing digital content is often easier than trudging to a retail store to buy a physical copy of a game. You're putting your faith in the platform holders to maintain their digital storefronts, the content on those storefronts, and their account systems so that your access keeps working. The recent closure of Nintendo's WiiU and 3DS e-shops was a stark reminder

that companies have the power to decide when you can buy digital content. While you can still re-download WiiU and 3DS games that you've purchased, it seems inevitable that Nintendo will stop letting you do that one day. It's already planning to shut down online services for those platforms after all, and remember when Google shut down Stadia. These recent PlayStation incidents are more aggravating, however, because of how sudden and seemingly unfair they are. With the discovery content, Sony is giving users a matter of weeks to watch their purchased shows for the last time before the shows are yanked from their library entirely. And Sony isn't offering any compensation for titles you've already bought or a way to transfer those purchases to another store. The PlayStation

account bans were as swift as they were unexpected, and while resolution for most arrived within a few hours, Sony still hasn't shared any public communication about what happened or why users should continue to trust the platform, end quote. Now, the answer here is obviously, if you can, if you can afford to do it, buy physical media when and where you can, actual disks and cartridges. But you've got to ask yourself, how much longer will companies even let us do that? Windows Central is reporting that Microsoft is planning to release the next major Windows version in 2024 with an AI-powered Windows shell, an advanced co-pilot, natural language search, and more. quote, with new leadership at the helm and X-Windows-Penay in the rearview mirror,

a new Windows roadmap is beginning to take shape under Microsoft's new Windows and Web Experiences team, which is now leading development on the next major Windows client update, codenamed Hudson Valley. We've known for some time that Microsoft is planning to ship a major new release of Windows platform in 2024 code named Germanium, which is what Hudson Valley will be based on. So what's the lowdown on this new Windows roadmap and what can we expect from the next version of Windows. Here's everything I know so far. According to my sources, the new Windows bosses are returning to an annual release cycle for major versions of the Windows platform, meaning Windows is going back to having just one big feature update a year instead of multiple

smaller ones throughout. Microsoft may still use Moment updates sparingly, but they will no longer be the primary delivery vehicle for new features going forward. Unsurprisingly, the big focus for Hudson Valley is on next generation AI experiences that are being woven and integrated throughout the OS, much of which will likely require new NPU hardware to function. According to my sources, Microsoft's Blockbuster new feature will be the introduction of an AI-powered Windows shell enhanced with an advanced co-pilot that's able to constantly work in the background to enhance search, jumpstart projects or workflows, understand context, and much more. Sources say these AI features will be groundbreaking. The company is working on a new history

timeline feature that will let users scroll back in time through all the apps and websites that Copilot has remembered, which can be filtered based on a user-specific search criteria. For example, you could type FY24 earnings, and every instance where that term was on screen will reappear for you to see an open. AI will also enhance search in Windows with the ability to use natural language to find things that you've previously opened or seen on your PC. If you don't remember the name or contents of a document, the search term, find me the document that Bob sent me on WhatsApp a few days ago is something search on Windows

will actually understand. Other AI features include something called Super Resolution, which will use NPU hardware to upscale the quality of videos and games. There's also an enhanced version of live captions in the works, which will be able to translate a number of different languages in real time, whether that be from audio or in a video or a live call. Microsoft is even working on AI-powered wallpapers, which will use machine learning to identify layers in any image and create a slight parallax effect that interacts with your

cursor or built-in gyroscope. if on a handheld device. Outside of AI, Microsoft wants to add a dedicated creator area to the start menu and File Explorer, which will congregate all of Microsoft services that let users create things in one place. This will essentially act as a launch pad for Microsoft 365, providing shortcuts to jump into a new or existing designer project, Word document, PowerPoint, presentation, etc. I also hear that Microsoft is working on several key improvements to energy saver, which sources say will be able to extend battery life by up to 50% on certain hardware. There's also a new green power feature in the works, which will try to charge your device when Windows detects

the power. It's pulling from the wall is renewable, end quote. Finally today, we know Apple has been working on something AI-related in the background, but it wasn't until yesterday that we got our first confirmation of that when Apple's machine learning research team quietly released MLX, an array framework to train and deploy ML models on Apple Silicon available on GitHub, quoting Computer World. In an X-note, Anni Hannon of Apple's ML team calls the software an efficient machine learning framework specifically designed for Apple Silicon, i.e., your laptop. The idea is that it streamlines training and deployment of machine learning models for researchers who use Apple hardware. MLX is a NUMPI-like array framework designed for efficient and flexible machine learning on Apple's processors.

This isn't a consumer-facing tool. It equips developers with what appears to be a powerful environment within which to build machine learning models. The company also seems to have worked to embrace the languages developers want to use rather than force the language on them, and it apparently invented powerful LLM tools in the process. Apple has provided a collection of examples of what MLX can do. These appear to confirm the company now has a highly efficient language model, powerful tools for image generation using stable diffusion and highly accurate speech recognition. This tallies with claims earlier this year and some speculation concerning infinite virtual world creation for future Vision Pro experiences.

Ultimately, Apple seems to want to democratize machine learning. Quote, MLX is designed by machine learning researchers for machine learning researchers, the team explains. In other words, Apple has recognized the need to build open, easy-to-use development environments for machine learning in order to nurture work further in that space. That MLX lives on Apple Silicon is also important, given that Apple's processors now live across all its products, including Mac, iPhone, and iPad, the use of the GPU, CPU, and conceivably at some point, neural engine on those.

chips could translate into an on-device execution of ML models for privacy with performance other processors cannot match, at least not in terms of edge devices, end quote. Listener at Kit Tsune, I don't know if I'm even remotely pronouncing that right, got in touch with me on X to show me that he had been testing that new Grock AI assistant from XAI, and when he prompted it to produce a script for a podcast, it spit out essentially my intro for this show, complete with calling it the TechMeme ride home. Can't confirm. Haven't tested this myself,

but the last link in the show notes is to his tweet with screenshots. Talk to you tomorrow.