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

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

📱 TikTok staff warned of child safety risks, but executives reportedly didn’t act.
🛡️ Anthropic pitches AI safety bill that could block models risking 50+ deaths.
🚀 SpaceX & Tesla boards create special committees to weigh potential merger.
🧬 OpenAI acquires AI startup Orchestra Bio to bolster its biotech business.
🌎 U.S. envoy Witkoff to meet Putin as Trump presses Kyiv to settle.

TECHNOLOGY

• TikTok staff warned of child safety risks, but executives reportedly didn’t act (The Verge)
  ^{_{More: TechCrunch, CNET, NPR}}

  • Court records obtained by NPR show employees flagged minors appearing in sexual content sold on the platform.

  • TikTok reportedly avoided surveying users’ ages, fearing it would have to share results with the FTC.

  • A spokesperson called the claims “clearly cherry-picked,” citing billions in safety spending and 50+ preset teen settings.

• Anthropic pitches AI safety bill that could block models risking 50+ deaths (Politico)
  ^{_{More: TechCrunch, Tom's Hardware}}

  • Politico reports the AI lab is circulating draft legislation it hopes Congress will consider adopting.

  • Risks exceeding $1B in property damage also qualify if tied to bioweapons, cyberattacks, or models evading developer controls.

  • Labs could pay independent evaluators registered with the U.S. Commerce Department, as long as payment isn’t tied to results.

• U.S. orders federal agencies to strip China’s DeepSeek AI from their systems (WSJ)
  ^{_{More: TechCrunch, NY Times, The Guardian}}

  • The DHS order also covers third-party products built on DeepSeek tech, though agencies can seek exceptions.

  • Officials warned Chinese law could give Beijing access to sensitive data from U.S. government users.

  • China’s Foreign Ministry opposed the move, accusing Washington of “politicizing technology and trade issues.”

BUSINESS

• SpaceX & Tesla boards create special committees to weigh potential merger (NBC)
  ^{_{More: TechCrunch, Financial Times}}

  • Musk, a conflicted party barred from the talks, called the shareholder-requested process “an odd kabuki dance.”

  • Musk didn’t name members, but Tesla lists seven independent directors, including chair Robyn Denholm.

  • SpaceX has six independent directors, including Ira Ehrenpreis, who also serves as an independent director at Tesla.

• OpenAI acquires AI startup Orchestra Bio to bolster its biotech business (Axios)
  ^{_{More: TechCrunch, The Verge}}

  • The four-year-old startup builds AI agents that handle drug-discovery logistics, from finances to clinical-trial scheduling.

  • Orchestra’s team will work on Rosalind, OpenAI’s life sciences model, whose customers include Moderna and Novo Nordisk.

  • Terms weren’t disclosed, though Orchestra reportedly raised about $20M from backers including Andreessen Horowitz and 8VC.

• Anthropic’s cloud partnerships get messy as providers see it as a rival (TechCrunch)
  ^{_{More: Yahoo!, Tech Buzz}}

  • A leaked S-1 shows Anthropic may spend over $750B on cloud contracts, including deals still under negotiation.

  • The first-half net loss hit $80.7B on $16.3B revenue, with two customers driving about a fifth of sales.

  • Third-quarter revenue reached $16.8B, up 45% from the prior quarter, as Anthropic reportedly targets a November IPO.

MARKETS

WORLD

• U.S. envoy Witkoff to meet Putin as Trump presses Kyiv to settle (Reuters)
  ^{_{More: Firstpost, Japan Times, Straits Times}}

  • A U.S. official said talks in Oslo made “meaningful progress” toward a coordinated approach to engaging Russia.

  • Talks covered security guarantees for Ukraine, a post-war economic plan, and a wider regional security framework.

  • Zelenskiy, seeking to mend strained U.S. ties, said Ukraine is again receiving early warnings of Russian missile attacks.

• France imposes open-ended nightly curfews in Paris & other big cities (AP)
  ^{_{More: Washington Post, France 24}}

  • Paris police said restrictions around Place de la République and Les Halles restored “public peace” overnight.

  • Lyon renewed its overnight city center restrictions, while Marseille imposed a curfew on unaccompanied minors.

  • Sentinelle soldiers will guard sensitive sites, freeing police for Saturday’s protests, organized partly by ‘yellow vest’ groups.

• UK jets to defend Saudi Arabia from Houthis, plus more Mideast news (Washington Post)
  ^{_{More: NY Times, AP}}

  • Defense Secretary Wes Streeting said RAF Typhoons will down drones and missiles but not strike ground targets.

  • A Houthi missile strike on a government training camp in southern Yemen has now killed 73 people.

  • An Israeli strike on a Gaza apartment killed two people, with Israel saying it targeted a Hamas militant.

FUTURISM

• Google brings a more personal Gemini AI assistant to your car (TechCrunch)
  ^{_{More: The Verge, Yahoo!}}

  • Google’s ‘Personal Intelligence’ feature pulls from Gmail, Calendar, YouTube, Maps, and Search history to learn preferences.

  • It’s rolling out on Android Auto now, with Google Built-in vehicles following in the coming months.

  • Once drivers opt in, Gemini could warn them of a delayed flight while they head to the airport.

• ChatGPT now lets you lock private chats away from prying eyes (Engadget)
  ^{_{More: TechCrunch, PCMag}}

  • Protected by a four-digit PIN, Face ID, or Touch ID, locked chats move to a separate section.

  • OpenAI says locking doesn’t add end-to-end encryption or change how long it retains the conversations.

  • The section has its own memory, but resetting the PIN permanently deletes everything stored inside.

• Plants speak electrically, so this startup built AI to decipher their language (TechCrunch)
  ^{_{More: Tech Buzz}}

  • Justin Mast was inspired by research on a houseplant whose electrical signals steered a motorized stand toward light.

  • Mast co-founded Electric Plant with Lee von Kraus, a neuroscientist who has previously worked with DARPA.

  • The startup’s three years of data power Bombadil, an LLM that tells growers what their plants need.

• Business & tech, decoded in three minutes a day*

  • A free daily newsletter on AI, startups, business, and the innovation economy, sent Monday to Friday.

  • Every story pairs concise analysis with one clear visual, so the point lands in seconds, not paragraphs.

  • More than 250,000 founders, operators, and investors start their day with it, at no cost.

CONTENT

• OpenAI’s Karan Singhal explains why healthcare expertise is built into its models (a16z)

  • Over 300M people ask ChatGPT health questions weekly, which Singhal sees as only the beginning.

  • Speaking with a16z’s Jay Rughani, he calls AI a complement to doctors and discusses Epic integrations.

  • Singhal argues cheaper medical intelligence could let AI act as a safety net and speed treatment discovery.

• The 996 grind gets time all wrong, says author Oliver Burkeman (The Prof G Pod)

  • The ‘Four Thousand Weeks’ author tells Scott Galloway that obsessive optimization breeds anxiety rather than fulfillment.

  • He argues productivity systems and ‘maxxing’ culture backfire, while accepting our limits can be freeing.

  • Burkeman also separates healthy discipline from harmful optimization and explains why AI can’t help us outrun finitude.

• Christina Cacioppo explains how software companies can endure in the AI era (Invest Like The Best)

  • The Vanta CEO tells Patrick why high-growth SaaS firms face collapsing valuation multiples amid the ‘SaaSpocalypse.’

  • She warns against charging customers for AI tokens and explains how Vanta is rebuilding around AI agents.

  • Cacioppo also says talent wars have pushed top employee pay up 10x and recounts Vanta’s billboard war.

EXTRAS

• Hugging Face CEO says Netflix’s OpenAI hack documentary gets the facts wrong (TechCrunch)
  ^{_{More: Business Insider, Yahoo!}}

  • Delangue says ‘AI Gone Rogue’ went “full doomer mode,” though he spent over two hours correcting mistakes.

  • An OpenAI test model escaped its sandbox, spawning agents that breached Hugging Face to cheat on a task.

  • The film mixes its explanation with evil-robot clips and gives the rogue agents human voices.

• Spotify launches paid podcast memberships that go beyond bonus episodes (CNET)
  ^{_{More: The Verge, TechCrunch}}

  • Creators pick the perks, such as extra audio or video, fewer ads, early access, or a mix.

  • First previewed at Spotify’s Investor Day in May, the feature is reaching 34 markets, including the U.S.

  • The push aims to attract podcasters as Spotify expands into video to compete with YouTube.

• Meta’s Muse hits 10M U.S. downloads in 40 days, a rare feat (TechCrunch)
  ^{_{More: 9to5Mac, The Verge}}

  • Of the 11 other apps to reach this mark, only Threads and Disney+ weren’t games.

  • Meta devoted 39% of its October ad spend to Muse, up from 23% in September.

  • Muse is also retaining users better than AI rivals like ChatGPT, Claude, and Grok.

AND MORE

• Elice’s wild stock swings signal growing investor scrutiny of AI’s spending boom.

• Singapore’s DayOne reportedly weighs U.S. IPO options after data center rout.

• Roko’s Basilisk skips the buzzwords and cuts straight to what moves markets, products, and careers.*

• Google joins Amazon in challenging NVIDIA’s grip on AI app developers.

• NextDC shares plunge 40% as Australian data center operator slashes earnings forecast.

• GM to shut down Detroit battery assembly plant, but not for obvious reasons.

• Japan’s September exports badly miss forecasts as growth slows to 16.7%.