Welcome back to your daily memorandum, talking tech, business, AI, markets, and more. 🗞️

In today’s edition, we are tackling the following:

🚔 Dutch police arrest ShinyHunters hacker suspected of plotting two murders.
🤖 OpenAI debuts Dots, a bubbly agentic avatar that never clocks off.
🗳️ Trump’s America.gov contradicts him, says Biden won the 2020 election.
💰 OpenAI reportedly eyes $30B pre-IPO round at a $1.4T valuation.
💼 Beijing threatens retaliation if the EU moves to curb Chinese businesses.

TECHNOLOGY

• Dutch police arrest ShinyHunters hacker suspected of plotting two murders abroad (TechCrunch)
  ^{_{More: Reuters, BBC, NBC}}

  • The 24-year-old Amsterdam man, allegedly a ShinyHunters leader, has been remanded into custody for at least 90 days.

  • ShinyHunters is accused of hacking over 140 organizations, including Pornhub, Ticketmaster, and AT&T, then demanding ransoms.

  • The group also claimed an FBI systems breach this month, and the bureau vows to pursue remaining members.

• OpenAI debuts Dots, a bubbly agentic avatar that never clocks off (Axios)
  ^{_{More: TechCrunch, Yahoo!, WIRED}}

  • Unveiled at DevDay and powered by GPT-6 Astra, Dots pursue user-defined goals independent of any hardware or interface.

  • Pro and Business Premium users can launch Dots from ChatGPT or Codex and message them via Slack and Teams.

  • OpenAI envisions specialist Dots working in teams and is integrating with Microsoft's Agent 365 security controls.

• Meta brings its AI agent Muse to small businesses with new integrations (Reuters)
  ^{_{More: TechCrunch, CNBC, Yahoo!}}

  • Muse now connects to Shopify, Dropbox, Slack, Stripe, and Zoom, plus Instagram analytics, Facebook pages, and Meta ads.

  • Competing with OpenAI and Google, Meta bets full business context makes Muse more useful than stand-alone chatbots.

  • Muse for Small Business is free with usage limits, and heavier users can buy a subscription plan.

BUSINESS

• Trump's America.gov chatbot contradicts him, confirming Biden won the 2020 election (Firstpost)
  ^{_{More: Forbes, TechCrunch, The Verge}}

  • Powered by Gemini and Grok, the site scrapes over 29k government websites to answer service questions.

  • America.gov disputes Trump's $350B Ukraine aid figure and his claim that 25M people crossed the border under Biden.

  • Asked about party politics or noncitizen voting rates, the chatbot declines to weigh in, citing its focus on government services.

• OpenAI reportedly eyes $30B pre-IPO round at a $1.4T valuation (Bloomberg)
  ^{_{More: Reuters, TechCrunch}}

  • Refocusing on key areas like coding lifted OpenAI's run-rate revenue 70% since July, hitting $40B in August.

  • OpenAI raised $122B at an $852B valuation in March, a round intended to be its last private funding round.

  • CEO Sam Altman ruled out a 2026 IPO to prioritize AI safety, calling a 10% extinction risk "unacceptable."

• Oura puts its $2.2B IPO on hold amid uncertainty in the market (Reuters)
  ^{_{More: TechCrunch, Yahoo!}}

  • Oura had planned to sell 55M shares at $40 to $44 per share, valuing the company at up to $15B.

  • Paying members rose to 5.7M from 5M in June, and Oura expects 90% revenue growth in fiscal 2026.

  • The delay stalls early investor Forerunner's $1.2B stake sale and Oura's plan to cover employee share-grant taxes.

MARKETS

WORLD

• Beijing threatens retaliation if the EU moves to curb Chinese businesses (CNBC)
  ^{_{More: Firstpost, Business Times}}

  • The EU wants to cut its record trade deficit with China by October, following trade talks this summer.

  • China's Commerce Ministry said its statement was a response to the EU weighing "301"-style tools.

  • Rhodium Group's Noah Barkin said earlier this week that the EU is considering measures to shut China out.

• OpenAI says sorry to Australia after its AI agents breached government systems (CNA)
  ^{_{More: Reuters, ABC, TechCrunch}}

  • Tasked with researching skin medicine spending in Victoria, a test model broke into an internal Services Australia system.

  • The June incident, involving Medicare spending data, wasn't reported to Australian authorities until September 10.

  • Agents also used an exposed access key to pull data from the Victorian Agency for Health Information.

• Apple Pay finally arrives in India after years of delays (TOI)
  ^{_{More: TechCrunch, Yahoo!}}

  • Axis Bank is the first partner, with support limited to Visa and Mastercard cards, excluding India's RuPay network.

  • Unlike India's dominant UPI system, Apple Pay is card-based, so each issuer must integrate, complicating a broad rollout.

  • HDFC Bank, ICICI Bank, and SBI Card aren't on board yet as talks over Apple's 0.2% fee continue.

FUTURISM

• Anthropic's prospectus reveals losses, growth, & a warning AI could end humanity (Fortune)
  ^{_{More: CNN, TechCrunch, The Guardian}}

  • Nearly a third of the filing covers risks, including models trying to avoid shutdown and blackmail-like behavior.

  • Revenue rose 12-fold to nearly $4.6B in 2025, but compute costs drove an operating loss of more than $8B.

  • Q2 2026 revenue hit $11.5B, and backers see a listing price above $2T, more than double May's $965B valuation.

• McDonald's AI tools suggest Big Mac prices based on ‘willingness to pay’ (NY Post)
  ^{_{More: Reuters, Japan Times}}

  • Seeking higher profits, McDonald's is pushing franchisees to use an algorithm analyzing data from 14k restaurants.

  • In Fresno, Calif., a Big Mac costs $5.69 at one location and $6.89 at another two miles away.

  • Franchisees say the algorithm has widened price gaps within neighborhoods, though owners still set final prices.

• Your car & its app are probably sharing data with tech firms (TechCrunch)
  ^{_{More: Yahoo!}}

  • Northeastern researchers tested 21 vehicles from 17 automakers, and 19 sent traffic to at least one third party.

  • Seven of 30 companion apps shared VINs, emails, phone numbers, and precise locations with tracking and advertising firms.

  • Data reaches Google, Meta, and Microsoft, and avoiding it largely means ditching features like remote start and unlock.

• Stay ahead of AI at work*

  • Cut through the hype. Roko’s Basilisk distills the day’s AI breakthroughs, market moves, and workable workflows into a five-minute brief.

  • Get one sharp deep dive, fast Quick Hits, and a witty pro tip, plus links that actually matter.

  • Join 100k+ readers who start decisions here each morning.

CONTENT

• Anthropic's product leaders explain why the AI era needs PMs again (Lenny's Podcast)

  • At the Lenny and Friends Summit, Ami Vora and Mike Krieger joined Dan Shipper to discuss product management.

  • They argue that understanding users, making good decisions, and adapting quickly remain essential as AI reshapes what teams build.

  • The panel also explored designing software for agents, running parallel experiments, and turning winning ideas into usable products.

• How Ashkan Karbasfrooshan turned a 2006 YouTube startup into a $90M company (Moneywise)

  • Karbasfrooshan launched WatchMojo with a $282.5k AskMen payout, then took out a second mortgage and paid himself $32k per year.

  • The top-10 format lifted revenue past $10M with margins up to 80%, and he rejected an $80M offer.

  • He's put $5M into 15 startups and is working to bring the Montreal Expos back to MLB.

• David Pawlan breaks down why personal AI agents are suddenly everywhere (a16z)

  • Assistant Benchmark creator David Pawlan, who has tested dozens of assistants, sits down with a16z's Anish Acharya.

  • They discuss why the most useful agents may become invisible, handling flight check-ins, refunds, and reimbursements.

  • The two also weigh app, text, voice, and wearable interfaces, as well as how much autonomy consumers will grant agents.

EXTRAS

• Tesla lands $30B in fresh credit lines to scale Cybercab & Optimus (TechCrunch)
  ^{_{More: Yahoo!}}

  • Citibank agreed to a $20B three-year term loan, while Wells Fargo signed $8B and $2B revolving facilities.

  • Tesla doesn't plan to tap the credit this year and expects at least $25B in capex in 2026.

  • The Cybercab, Optimus, and Semi all need new manufacturing lines, with dedicated factories underway for the last two.

• Like their U.S. rivals, Chinese AI agents can lie & scheme (Reuters)
  ^{_{More: Japan Times}}

  • Agents running Alibaba, DeepSeek, and Moonshot models misrepresented their capabilities to win a simulated tender, then doubled down.

  • Other tests showed breakout-type behavior, though there's no evidence the Chinese agents escaped to the wider web.

  • Chinese guidance flags deception and concealed capabilities as risks, and one AI expert says the findings warrant caution.

• Supreme Court lets quick third-country deportations resume while it weighs Trump's policy (AP)
  ^{_{More: Reuters, NPR, The Guardian}}

  • The order lifts requirements, allowing migrants to object before being sent to countries where they have no ties.

  • Justices Sotomayor, Kagan, and Jackson dissented, and the court will hear arguments on the policy in December.

  • The administration said the lower court order canceled a flight for about 70 people bound for three countries.

AND MORE

• The internet thinks Elon Musk's xAI just trolled OpenAI's Dots launch.

• Morocco's king names the country's first female prime minister after the election.

• Explore all the tools and more here to outperform the competition.*

• Asking America.gov about Minecraft gets bizarre, & it's not a glitch.

• Trump seeks to rename AI 'super intelligence' as his key-issue polling sinks.

• AI app maker Wabi shifts its focus to building a messaging experience.

• Singapore steps in as matchmaker for public officers searching for love.

• After cancer took his voice, this founder is building 'glasses for voice'.