☕️ Apple is cooking up a trio of AI wearables

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

👓 Apple is cooking up a trio of AI wearables

💰 Nvidia to sell Meta millions of chips in multiyear deal

🤖 Anthropic launches Claude Sonnet 4.6 as new default model

🔍 Perplexity drops ads from its AI search engine

🚕 First Tesla Cybercab rolls off Giga Texas line

🐛 Microsoft bug let Copilot access confidential emails

🎵 Google Gemini can now generate music with Lyria 3

Plus: 🎁 19 other news you might like, 🧰 6 tools, and 📚 4 papers.

👓 Apple is cooking up a trio of AI wearables LINK

Apple is speeding up work on three AI wearables: a camera-equipped pendant similar to an AirTag, smart glasses code-named N50, and AirPods with new AI features, according to a Bloomberg report.

The smart glasses will reportedly include a high-resolution camera, with Apple targeting the start of production as early as December and a public release in 2027, putting it against Meta and Snap.

All three devices are designed to connect to the iPhone and will include Siri as a critical part of the experience, with the glasses described as more upscale and feature-rich than the others.

💰 Nvidia to sell Meta millions of chips in multiyear deal LINK

Meta announced a deal worth tens of billions of dollars to buy millions of Nvidia AI chips, including standalone CPUs and next-generation Vera Rubin systems, for its data centers.

Meta becomes the first company to deploy Nvidia's Grace central processing units as standalone chips in data centers, rather than pairing them alongside GPUs in a server.

Meta plans to spend up to $135 billion on AI in 2026, and analyst Ben Bajarin said a good portion of that capital expenditure will go toward this Nvidia build-out.

🤖 Anthropic launches Claude Sonnet 4.6 as new default model LINK

Anthropic released Claude Sonnet 4.6 as the new default model for Free and Pro plan users, continuing the company's pattern of updating its mid-size Sonnet model roughly every four months.

The beta release includes a 1 million token context window, double the previous largest Sonnet window, which Anthropic says can hold entire codebases, lengthy contracts, or dozens of research papers.

Sonnet 4.6 posted record benchmark scores in computer use and software engineering, plus a 60.4% on ARC-AGI-2, though it still trails Opus 4.6, Gemini 3 Deep Think, and a refined GPT 5.2.

🔍 Perplexity drops ads from its AI search engine LINK

Perplexity, the AI search startup that was among the first AI companies to introduce ads, is now dropping advertising from its product and has no plans to pursue it further.

A Perplexity executive told the Financial Times that ads made users "start doubting everything," undermining trust in the chatbot's answers and making people less willing to pay for it.

The move comes as competitors head the other direction: OpenAI recently started testing ads on ChatGPT, Google shows ads in AI search results, while Anthropic has committed to keeping Claude ad-free.

🚕 First Tesla Cybercab rolls off Giga Texas line LINK

Tesla has built the first Cybercab at Gigafactory Texas, hitting a milestone before the autonomous two-seater's planned production start in April, according to a post from the company on X.

The Cybercab has no pedals or steering wheel and will rely entirely on Tesla's vision-based Full Self-Driving system, positioning it to compete with autonomous services like Waymo.

Musk warned that early production will be "agonizingly slow" because almost everything in the Cybercab is new, following an S-curve before volumes eventually ramp up significantly.

🐛 Microsoft bug let Copilot access confidential emails LINK

Microsoft confirmed a bug in its Copilot AI let the tool read and summarize confidential emails for weeks, even when customers had data loss prevention policies in place to block that.

The issue, tracked as CW1226324, affected Copilot Chat in Microsoft 365 since January, incorrectly processing draft and sent messages that had a confidential label applied to them.

Microsoft began rolling out a fix in February but did not say how many customers were affected, while the European Parliament separately blocked built-in AI features on work devices over privacy concerns.

🎵 Google Gemini can now generate music with Lyria 3 LINK

Google is rolling out a music-generation feature in the Gemini app, powered by DeepMind's Lyria 3 model, which can create 30-second tracks with lyrics and cover art from text prompts.

Users can upload a photo or video to generate a matching song, control elements like style, vocals, and tempo, and all output carries a SynthID watermark to identify AI-generated content.

Google is also expanding Dream Track on YouTube globally for creators, though the tool won't mimic artists directly — naming one in a prompt only inspires a track with similar style or mood.

Other news you might like

• Amazon quietly cancels Blue Jay warehouse robot months after debutLINK

• Tesla dodges 30-day suspension in California after removing AutopilotLINK

• Palantir moving its headquarters from Denver to MiamiLINK

• WordPress.com adds an AI Assistant that can edit, adjust styles, create images, and moreLINK

• Stripe's stablecoin firm Bridge wins initial approval of national bank trust charterLINK

• Chinese hackers exploited a Dell zero-day for 18 months before anyone noticedLINK

• More than 50% of enterprise software could switch to AI, Mistral CEO saysLINK

• U.S. court bars OpenAI from using ‘Cameo’LINK

• Uber to invest over $100 million in autonomous vehicle charging amid robotaxi pushLINK

• Running AI models is turning into a memory gameLINK

• Password managers' promise that they can't see your vaults isn't always trueLINK

• Waymo defends use of remote assistance workers in robotaxi operationsLINK

• European Parliament blocks AI on lawmakers’ devices, citing security risksLINK

• Amazon has lost $450 billion in value during this historic losing streakLINK

• Tesla Robotaxis Reportedly Crashing at a Rate That’s 4x Higher Than HumansLINK

• Mark Zuckerberg to testify in social media addiction trialLINK

• Snap's direct revenue hits $1 billion annualised run rate as subscribers top 25 millionLINK

🧰 Trending tools

Sonnet 4.6: a conversational AI assistant that handles coding, analysis, writing, and research tasks with improved reasoning capabilities and context understanding.LINK

Moda: an AI-powered design tool that generates fully editable, brand-aligned content on a layered canvas for slides, posters, and ads.LINK

Omnia: a monitoring tool that reveals how AI models perceive and cite your brand, enabling you to optimize content for better AI search visibility.LINK

AsteroidOS 2.0: an open-source Linux distribution for smartwatches, providing an alternative operating system to proprietary wearable platforms.LINK

Flixier Generate AI Video in Timeline: generates and assembles AI video clips directly in your editing timeline, letting you create, extend, and polish videos with voice and captions in one workspace.LINK

Design Rails: generates complete brand identity packages (logo, colors, typography, UI styles) as downloadable files that AI coding agents can use to build consistent, branded interfaces.LINK

📚 Trending papers & reports

GLM-5 AI model matches GPT-4 performance: the Chinese system achieves similar accuracy on reasoning and language tasks while using fewer computing resources during training.LINK

Robots learn faster by remembering past attempts: instead of starting fresh each time, the system stores what worked before and reuses those solutions for similar new tasks.LINK

Compose Verifiable Prompts for Reinforcement Learning: training language models by breaking complex tasks into smaller, checkable steps improves accuracy because each piece can be verified before moving forward.LINK

Quantum RSA Breaking Gets Cheaper: researchers reduced the number of quantum bits needed to crack RSA 2048 encryption from millions down to 100,000 physical qubits.LINK

Want to get the latest news differently? Find us on:

See you tomorrow for a new dose of ☕️ Techpresso!