☕️ Stripe bids $53B to buy PayPal

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

💳 Stripe bids $53B to buy PayPal

🔊 OpenAI's first device is screen-free

🍎 China approves Apple Intelligence

🇬🇧 UK plans teen social media curfew

🦅 White House launches AI cyber defense

⚖️ Meta used AI to target disabled workers for layoffs, lawsuit claims

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

💳 Stripe bids $53B to buy PayPal LINK

Stripe and private equity firm Advent International have offered $60.50 a share in cash for PayPal, a bid valuing the payments company at over $53 billion and backed by roughly $50 billion in bank financing.

The two buyers would own equal stakes and, unusually for a deal involving private equity, say they would not break PayPal up, with the price sitting about 28% above the stock's recent close near $47.

Stripe, valued at $159 billion and processing $1.9 trillion in payments last year, would gain a consumer brand people already trust with their card numbers, while Advent would end up funding rivals on both sides.

🔊 OpenAI's first device is screen-free LINK

OpenAI is reportedly building its first consumer hardware device, a screen-free home companion centered on ChatGPT that Bloomberg says is described internally as a "humanlike AI companion that lives in the home."

Instead of just answering voice commands like a normal smart speaker, the device is meant to learn about its owner over time and pull from their digital life, including emails, while including moving mechanical parts to feel more like a companion.

Former Apple engineers who worked on the iPhone and Mac have reportedly joined the project, which comes as Apple sued OpenAI last week, claiming a former employee took confidential files about unreleased products, allegations OpenAI denies.

🍎 China approves Apple Intelligence LINK

China's internet regulator, the Cyberspace Administration of China, has cleared Apple Intelligence, the set of AI tools that summarize emails, write reports, and edit photos, for use on iPhones sold in the country.

The approval covers only iPhones, and the official notice did not say whether iPads or Mac computers sold in China would also be allowed to run the AI features; Apple has not commented.

Apple Intelligence and Samsung's Galaxy AI were the only foreign services approved among seven, with local vendors Huawei, Oppo, Vivo, Xiaomi, and ZTE also cleared, while Baidu said it is building Apple's China features.

🇬🇧 UK plans teen social media curfew LINK

The UK will make social media companies automatically block users aged 16 and 17 from platforms between midnight and 6 am, a curfew feature teens can choose to switch off themselves.

The optional overnight lockout comes with a full ban on social media for children under 16, expected in spring 2027, and turns off autoplaying videos and personalized feeds by default for older teenagers.

The Department for Science, Innovation & Technology also plans mandatory chatbot breaks for under-18s, bans on AI that fakes romantic talk, and updated school lessons on AI, bias, and disinformation.

🦅 White House launches AI cyber defense LINK

The White House has created a new group called "Gold Eagle" that unites AI developers and critical service providers to strengthen national cybersecurity by sharing information about software flaws found by AI systems.

Set up after an executive order President Trump signed on June 2, the group includes the White House, Homeland Security, the Treasury, the Department of War, and private companies handling US critical infrastructure and open-source software.

Gold Eagle has begun receiving and ranking vulnerability reports across industries, using AI to cut down duplicate scanning and quickly pass threat and repair details to those defending federal and private sector systems.

⚖️ Meta used AI to target disabled workers for layoffs, lawsuit claims LINK

Twenty-six current and former Meta workers have sued the company in a Northern California federal court, claiming it used AI tools to help pick who lost their jobs in the roughly 8,000 layoffs in May.

The suit says Meta scored and ranked staff using systems like its Metamate chatbot, AI usage dashboards, and keystroke-monitoring data, marking lower output during medical, parental, or disability leave as underperformance rather than adjusting for the time away.

The plaintiffs want a judge to pause their firings, order an independent audit, and recalculate the selections without counting protected leave, while Meta says the claims "lack merit" and that people, not AI, made the decisions.

Other news you might like

• Intel is the first chipmaker to mass-produce chips using ASML's next-gen printing toolLINK

• OpenAI’s new flagship model deletes files on its own, people keep warningLINK

• ASML is raising its annual sales forecast for the second time this year on surging AI chip demandLINK

• Google and Epic give up fighting — third-party Android app stores are coming next weekLINK

• Introducing Claude for TeachersLINK

• Meta’s Adam Mosseri says AI token budgets could soon be capped per engineerLINK

• Google faces another AI training lawsuit from major publishersLINK

• Jamie Dimon says JPMorgan has slashed 40% of jobs in some departments, thanks to AILINK

• Massive leak reveals everything you want to know about the Galaxy Z Fold 8 and Fold 8 UltraLINK

• Uber (UBER) Stock Slides as €10 Billion Delivery Hero Acquisition Nears CompletionLINK

• Spotify now lets Premium users talk to a chatbot about music, podcasts, and booksLINK

• Anthropic’s newest ad is creeping people outLINK

• Google revamps image search for its 25th anniversary with more images and more AILINK

• You Just Hired a Million Bad EmployeesLINK

🧰 Trending tools

QuickQuill: a text expansion tool that turns short snippets into full paragraphs using smart templates and shortcuts, saving hours on repetitive writing tasksLINK

RecordMeeting: records calls from Google Meet, Zoom, Teams, Webex, Telegram, WhatsApp, and Discord, then auto-transcribes them into searchable notes with speaker identification and action items.LINK

Crustdata Recruiter: provides APIs to search and enrich company and people data, tracking real-time events like job changes, promotions, or new posts.LINK

YAGNI: a platform for managing AI agent teams that earn autonomy over time, letting you assign tasks, track performance, and retain control over critical decisions.LINK

nudge2.0: turns brain-dumped tasks into an auto-scheduled weekly calendar, using money-backed deadlines and Slack/Discord alerts to enforce follow-through.LINK

📚 Trending papers & reports

Long computer tasks still trip up top AI agents, with the best system finishing only 20.6% of realistic hour-plus workflows despite hundreds of tool calls.LINK

AI test scores often understate how good chatbots really are, since letting them use more computing time and more attempts sharply boosts results on hard tasks like coding, math, and cybersecurity.LINK

Chess move prediction gets a boost from Matilda, a small add-on model that predicts elite players' moves far better, up to 21.9% more accurately above 3000 Elo, while barely changing lower-rated predictions.LINK

Spreadsheet automation using plain-English commands got a big accuracy boost, hitting 100% task completion and 54.8% correctness, beating a rival tool's 44.3%.LINK

Memory conflicts in language models found that mixing learned facts with new context causes confident wrong answers, but measuring the model's internal "distance" to known facts spots errors better than checking its confidence, with zero false refusals.LINK

See you tomorrow for a new dose of ☕️ Techpresso!