☕️ US tech giants back open-weights AI

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

🤖 US tech giants back open-weights AI

🩺 ChatGPT Health opens to all US adults

💳 Stripe eyes $10B OpenRouter deal

🤳 Facebook adds selfie verification badge

🍎 Apple presses suppliers on 2026 iPhone prices

🗣️ Claude voice mode gains smarter models

Plus: 🎁 15 other news you might like, 🧰 5 tools, and 📚 5 papers.

🤖 US tech giants back open-weights AI LINK

Major American technology companies have publicly endorsed open-weights AI models, a stance that Nvidia founder Jensen Huang highlighted by sharing a joint letter in his first-ever post on X.

The firms argue that open-weights AI, where the model's trained settings are made freely available, is important for the future of the technology, though the article gives few specifics on which companies signed.

Huang's use of a debut X post to spread the letter drew attention to the effort, signaling that leading US players want to promote the open approach rather than keep AI models closed.

🩺 ChatGPT Health opens to all US adults LINK

OpenAI is opening ChatGPT Health, its tool for answering health questions, to every U.S. user over 18 across all plans, arriving this week for logged-in people on free, Go, Plus, and Pro tiers via web and iOS.

The feature lets users link medical records and personal data from services like Apple Health, MyFitnessPal, Epic, and Oracle Health, and now works in any chat rather than only inside a dedicated hub, where 70% of health queries happened outside anyway.

The rollout came a day after a Florida pastor sued OpenAI over a near-fatal suggestion to skip seeing a doctor, and the company still says its services are not meant to diagnose or treat conditions, urging users to check with professionals.

💳 Stripe eyes $10B OpenRouter deal LINK

Stripe is in talks to buy OpenRouter, an AI startup that connects developers to many different models, in a deal that could reach roughly $10 billion, according to The Wall Street Journal.

OpenRouter runs a marketplace listing hundreds of large language models from OpenAI, Anthropic and open-weight options, letting customers compare, switch, and control spending while avoiding reliance on a single AI provider.

The startup was valued at only $1.3 billion in May, so a $10 billion sale marks a sharp jump in months; its backers include Menlo Ventures and Alphabet's CapitalG, and it already uses Stripe for payments.

🤳 Facebook adds selfie verification badge LINK

Facebook is rolling out a free Verified program that uses a facial recognition selfie, matched against your existing profile photos, to confirm that a real person is behind an account in a process Meta says takes only a few minutes.

To qualify you must be over 18 and in good standing with Facebook's community standards, after which the badge appears across Marketplace, Dating and other parts of the service as identity confirmation only, not an endorsement or trust guarantee.

Meta is launching this as scammers use generative AI to build fake profiles with realistic images and stories that send automated messages at scale, though the badge won't cover Pages or ProMode accounts and will start in select markets before expanding.

🍎 Apple presses suppliers on 2026 iPhone prices LINK

Apple is pushing its display suppliers to accept lower prices for the new iPhone screens as a way to offset the sharply rising cost of memory chips during the current RAM shortage.

The company got Samsung Display and LG Display to drop iPhone 18 Pro Max panel prices to about $66.50, down from $80 for the 17 Pro Max and over $100 for some 16 Pro Max screens.

The price cut is notable because the new OLED screens use M16 stacks for better brightness, color, power use, and lifespan, which makes them harder to produce and forces makers to adjust their process and stabilize yields.

🗣️ Claude voice mode gains smarter models LINK

Anthropic is upgrading Claude's voice mode so the model picker actually matters, with Opus and Sonnet now running spoken conversations instead of quietly falling back to the older Haiku 4.5 model behind the scenes.

The voice picker offers Opus, Sonnet, and Haiku plus a control for setting the thinking level, though the newer Fable model is missing, and spoken replies still appear to run through ElevenLabs rather than a speech-native system.

Putting stronger reasoning models into voice suits longer thinking and tool-heavy spoken requests, and in testing the mode paused when interrupted and waited through long silences, marking a different path from OpenAI's speech-native approach.

Other news you might like

• Intel Q2 sees ‘strongest revenue growth in more than 15 years’LINK

• Amazon is requiring sellers to label AI-generated people in product images after N.Y. lawLINK

• Pentagon awards Oracle nearly $7 billion deal in latest software consolidation pushLINK

• Microsoft launches new in-house AI models it says cut costs up to 89% versus OpenAILINK

• Coinbase lets businesses accept USDC payments from AI agentsLINK

• Google’s anti-search-scraping lawsuit dismissedLINK

• Wise shares tumble after US regulator rejects its trust bank charter bidLINK

• AMD takes on Nvidia with its Helios AI rack scale systemLINK

• Facebook launches a dedicated Marketplace app for sellers, adds a free verification systemLINK

• Google now owns 5% of SpaceX after historic IPOLINK

• Midjourney acquired the astrology app Co-StarLINK

• Black Forest Labs launches FLUX 3 capable of generating images and 20-second video with audio — but in limited release to startLINK

• TikTok faces massive EU fines over online child safety concernsLINK

• Amazon cuts jobs in AGI unit as it sharpens focus on key AI projectsLINK

🧰 Trending tools

Buzz: I don't have enough information about what Buzz actually does. The description "Come test the early stages with us" tells me it's in early testing but doesn't explain its functionality, purpose, or the problem it solves. Could you share more details about: - What category of tool this is (productivity, communication, dev tool, etc.) - What specific problem it solves - Key features it offers With that information, I can write an accurate one-line summary that meets all the format requirements.LINK

Liso: converts highlighted web text—articles, newsletters, posts—into instant audio, letting you listen instead of read when you're busy.LINK

MinkNote: a local-first macOS notes app storing entries as plain Markdown files, with projects, tags, fast search, and no account required.LINK

Hotspot Meter: tracks data usage per Wi-Fi network or hotspot from your Mac's menu bar, showing live speeds, historical totals, and cap-alert warnings, all processed locally with no accounts or tracking.LINK

Freesolo Flash: lets enterprise teams train small, specialized models via reinforcement learning, turning generic AI capability into product-specific features without deep ML expertise.LINK

📚 Trending papers & reports

Hidden bias testing finds chatbots discriminate over six times more when demographic clues are implied rather than stated outright, with caste bias so severe that even the best fix leaves it four times worse than other categories.LINK

AI double agents that role-play as decoys, feeding attackers fake confirmations instead of real secrets, can be trained via reinforcement learning to fool even strong models like Gemini3-Pro and GPT-5.4, offering a new defense against chatbot data-extraction attacks.LINK

Robot training habits tested across 100 real-world runs on three robot types show that some standard reinforcement-learning defaults actually hurt performance, while a simpler proven setup makes robots learn reliably with far less engineering effort.LINK

AI preference training often accidentally weakens the good response while trying to suppress the bad one, and a new rebalancing fix corrects this, improving downstream chatbot performance across multiple test settings.LINK

Robot memory compression lets robots condense long task histories into fixed-size summaries, so they handle jobs requiring long-term memory without the usual computational costs or overfitting failures of simply widening the context window.LINK

See you tomorrow for a new dose of ☕️ Techpresso!