A New Mexico judge ordered Meta to pay $25 billion over the Cambridge Analytica scandal, the largest penalty of its kind the company has faced. Meta says it will appeal, arguing the ruling makes it nearly impossible to do business in the state.

Today we are covering:

• Meta ordered to pay $25 billion in New Mexico over the Cambridge Analytica scandal

• Pentagon quietly builds a Technology Support Office to covertly disrupt foreign tech programs

• Hackers hit two critical PHP flaws days before the patch, with attempts climbing into the millions

• Meta beats Q3 estimates with $64.6 billion in revenue as Muse tops 10 million accounts

• Apple adds a Made for iPhone Duo badge to the App Store ahead of launch

Let’s dive in

Meta ordered to pay $25 billion penalty in New Mexico Cambridge Analytica case
By Jonathan Vanian via CNBC

• A New Mexico judge ordered Meta to pay $25 billion for violating the state’s consumer protection laws. The state had asked for between $35 billion and $45 billion, and the money will go to New Mexico’s education fund.

• The ruling follows a September jury verdict that found Meta liable for more than 43 million violations for misleading Facebook users about privacy and content moderation. The case centers on a personality quiz app that collected data from about 87 million profiles, which was then sold to Cambridge Analytica.

• Meta says it will appeal, and its chief legal officer, C.J. Mahoney, argued the approach “makes it all but impossible to do business in the state.” Attorney General Raúl Torrez called it one of the largest state verdicts ever against the company.

𝕏: Ka-boom. As I’ve been saying and dissecting for the last eight years, there is a lot more there there. Every settlement has been in part due to the increase in temperature when you get to the top of Meta and its board. Good on New Mexico. - Jason Kint (@jason_kint)

The Pentagon is quietly building an elite new tech sabotage agency
By John Sakellariadis, Jack Detsch and Paul McLeary via Politico

• The Pentagon has created a Technology Support Office to covertly disrupt and sabotage foreign technology programs seen as threats to U.S. military superiority. Deputy Defense Secretary Steve Feinberg signed the memo establishing it on April 14.

• It was built by reorganizing two classified intelligence programs that have run similar efforts since 2023. Its tools reportedly include custom hacking capabilities and deception operations meant to trick adversaries into installing faulty components in weapons platforms.

• The budget is not public, but a former official estimates its classified contracting could top $1.5 billion. Critics worry about secrecy, thin congressional briefings, and Feinberg’s former private equity firm’s defense investments, while the Pentagon calls the conflict concerns “a baseless, completely recycled narrative.”

𝕏: New: I got a tip a little while ago about a shadowy Pentagon entity called “TSO” with a lot of money and a desire for offensive hacking tools. Turns out TSO is a new DoD agency. And it does a lot more than hacking. - John Sakellariadis (@johnnysaks130)

The best way to reach new readers is through word of mouth. If you click THIS LINK in your inbox, it’ll create an easy-to-send pre-written email you can just fire off to some friends.

Hackers Exploited the PHP Vulnerabilities Days Before They Were Patched
By Lily Hay Newman via Wired

• Attackers went after two critical PHP flaws, nicknamed PHP2Shell (CVE-2026-109618 and CVE-2026-109619), before patches were available. Amazon spotted an attempt on its honeypot on October 17, four days before disclosure, and details leaked publicly on October 20.

• The bugs allow remote code execution without a login on PHP versions going back about a decade, and PHP runs nearly 70% of sites whose server language is known, including WordPress, Drupal and Joomla. Fixes landed in PHP 8.2.35, 8.3.36, 8.4.27 and 8.5.12.

• Cloudflare CEO Matthew Prince said the company saw thousands of attempts in the 10 hours before disclosure, rising to millions soon after. PHP 8.1 and older get no official patch, and it is still unclear whether any attempts succeeded.

𝕏: Fascinating. Amazon says it has telemetry showing PHP2Shell was being exploited on Oct. 17—that is, four days prior to public disclosure. Not known who was doing it. “It remains unclear how the actor acquired exploit information prior to disclosure.” - Dustin Volz (@dnvolz)

Meta CEO Mark Zuckerberg says 10 million people have set up a Muse AI agent
By Natalie Musumeci via Business Insider

• Meta reported third-quarter revenue of $64.6 billion, up 26% and above the $63.34 billion analysts expected. Net income was $9.86 billion, or $3.83 per share, after a $9.89 billion charge tied to a legal settlement.

• Mark Zuckerberg said more than 10 million people in the U.S. and Canada have set up accounts for Muse, Meta’s personal AI agent launched in August. Meta also introduced Muse Wonder, which it calls its first frontier model.

• Meta forecast fourth-quarter revenue of $73 billion to $76 billion, a midpoint above the $73.65 billion analysts expected. Its apps averaged 3.64 billion daily users, headcount fell 15% to 66,775, and shares rose 2.9% after hours.

𝕏: Muse, the most expensive toy ever by Meta (and some smart people may ask why it reminds of Manus AI), is draining all of Meta’s compute, and they are gonna get more. The problem? Many who downloaded it and tried it have already abandoned it. There is no clear visibility on user retention for this. - Kakashii (@kakashiii111)

Apple will label apps that are “Made for iPhone Duo”
By Emma Roth via The Verge

• Apple will add a Made for iPhone Duo badge to App Store product pages for apps optimized for its foldable, and plans curated collections of iPhone Duo apps and games. The badge arrives a day before the device goes on sale on October 23.

• Apps fall into tiers: those built for the foldable, apps optimized for iOS 27 that adapt automatically on the inner screen, and regular iPhone apps that behave like iPhone apps on an iPad. Netflix, Prime Video, Word, Outlook, Zoom, Slack, Salesforce and Fantastical are highlighted for the best experience, though optimized apps are a minority at launch.

• The iPhone Duo starts at $1,999, and Apple says it is sold out through the end of January. Susan Prescott, Apple’s vice president of Worldwide Developer Relations, said developers are “reimagining what their apps can do.”

𝕏: The iPhone Duo reviews are some of the most brutal I’ve seen from an Apple product in a long time (Vision Pro aside). But I’ve been around long enough to remember the iPhone 5, iPhone 6 Plus and iPhone X. The app optimization issue is a non-issue and will be resolved in weeks. - Mark Gurman (@markgurman)

Trending in AI

Today’s big AI theme: AI agents are getting real jobs, from proactive assistants and online shopping to lab research.

• Anthropic’s answer to Muse is here: Claude Duet, a mobile and desktop AI agent that takes actions proactively

• OpenAI unleashed thousands of AI agents to study a Parkinson-like disease — here’s what they found in 32 minutes

• Hugging Face’s New AI Biology Model Lets Users Chat With DNA

• While Amazon blocks AI agents, Walmart rolls out the welcome mat

• Empires of AI: The Rise and Fall of a New Global Order

Thanks for reading to the bottom and soaking in our Newslit Daily, fueled with highlights for your morning.

I hope you found it interesting and, needless to say, if you have any questions or feedback, let me know by hitting reply.

Take care and see you tomorrow!

— José Montes de Oca

How was today’s email?

🫤 Not Great | 🙂 Good | 🤩 Amazing

Apple’s first foldable is here, and the reviews say it is a beautiful, expensive first draft: The Verge liked the matte inner screen but found the iPhone Duo runs hot and may not last a full day of heavy use.

Today we are covering:

• iPhone Duo reviews are in: Apple’s first foldable impresses on design but runs hot at $1,999

• OpenAI brings its GPT-6 models to Google Cloud, expanding beyond Microsoft

• Two critical PHP flaws trigger a rush to patch ahead of an emergency release today

Read more

A power failure in Google Cloud’s London region knocked out Revolut’s cards and app early Tuesday and disrupted other UK banks. It is a fresh reminder of how much everyday banking depends on a handful of cloud regions.

Today we are covering:

• Google Cloud outage in London knocks out Revolut and disrupts other UK banks

• Developers use AI agents like Claude Code to strip ads and tracking from smart TVs

• Documents suggest the Pentagon wants an agentic AI framework called ECHOTRIBBLE to hack adversaries

Read more

It’s Monday, October 19. Locked in and ready to go. Anthropic is putting 5% of its shares into a new public benefit fund ahead of its expected IPO. The fund is a company-owned LLC aimed partly at AI-driven job disruption, and Elon Musk has already offered a rival idea.

Today we are covering:

• Anthropic commits 5% of its shares to a new public benefit fund ahead of its expected IPO

• Apple is reportedly working on a HomeView with a robotic arm that turns toward whoever is speaking

• An X product leader says Threads and Bluesky have “quarantined” some of the worst people online

Read more

It’s Friday, October 16. Friday is here — close the tabs, step away from the screen, and enjoy the break.

Elon Musk showed a Tesla Roadster that launched off a ramp using SpaceX hardware, rolled in the air, and landed on its wheels. Production is not planned until 2028, and Musk himself says sales will not move the needle financially.

Today we are covering:

• Elon Musk unveils a Tesla Roadster that “flies” with SpaceX hardware, with production planned for 2028

• Meta releases Muse Spark Open, an 816B-parameter open-weight model that runs on two Mac Studios

• Meta’s new commercial terms could let it scrape news sites, legal experts say

Read more

Citizen Lab says an Israeli ad company turned ordinary targeted ads into a delivery system for zero-click spyware, and Google has already cut it off. It is the first documented case in the wild of phones being infected this way, with no click from the target.

Today we are covering:

• Citizen Lab exposes AdFection, a campaign that used targeted ads to deliver zero-click spyware

• A federal appeals court lets Florida enforce its law barring children under 14 from social media accounts

• Ledger unveils the Atom, an Android companion device, while it investigates tampered wallets

• Flock CEO tells TechCrunch Disrupt that “we shouldn’t pause progress” on surveillance

• Nvidia’s RTX Spark inside the Surface Laptop Ultra beats an M5 Max MacBook Pro in local AI tests

Let’s dive in

‘Nightmare Scenario’: Zero-Click Spyware Is Infecting Phones via Targeted Ads
By Matt Burgess via WIRED

• Citizen Lab published research on AdFection, which it calls the first in-the-wild demonstration of zero-click mercenary spyware delivered through targeted ads. An Israeli company, Discover-Tech, ran the ad platform, and devices at Al Jazeera were infected with Candiru spyware.

• When an ad loads, the user’s ad identifier and IP address go to Discover-Tech’s servers. If the person is selected as a target, the device receives malicious JavaScript that exploits it, and the ads themselves carry no exploit code, which helped them avoid detection by ad platforms.

• Discover-Tech ads sent millions of devices in dozens of countries to its servers, including networks tied to governments in 24 countries. Google and Beeswax cut its access, and IAB Europe removed it from its vendor list, while Citizen Lab says industry self-regulation failed to catch it.

𝕏: NEW @citizenlab report We uncovered an Israeli firm using online ads to run vast spy operation This is the first-ever public confirmation of zero-click mercenary spyware delivered via targeted ads. 45 devices at @AJEnglish confirmed infected - profdeibert (@RonDeibert)

Florida can enforce state law barring young kids from social media, appeals court rules
By Diana Novak Jones via Reuters

• The 11th Circuit reversed the injunction that had blocked Florida’s HB 3, so the state can enforce a law that bars children under 14 from holding accounts on platforms with addictive features and requires parental consent for 14- and 15-year-olds.

• Judge Kevin Newsom wrote that the law implicates the First Amendment but likely survives intermediate scrutiny because it is content-neutral, and he described it as a “conditional ban” aimed at features like infinite scroll, push notifications and auto-play.

• The tech groups CCIA and NetChoice, whose members include Meta’s Facebook and Instagram, Google, Snapchat and Pinterest, challenged the law. NetChoice says it is evaluating all of its options, and the litigation continues.

𝕏: Not good news, but grateful that @NetChoice and @Paul_Taske will continue this fight. To ban under-14s the platforms have to know the age of every user. To get parent permission for 15 and 16s will require some way to link the parent and child relationship… - Julie Barrett (@juliecbarrett)

The best way to reach new readers is through word of mouth. If you click THIS LINK in your inbox, it’ll create an easy-to-send pre-written email you can just fire off to some friends.

Ledger’s new hardware wallet is an Android phone without the phone
By Dominic Preston via The Verge

• Ledger unveiled the Atom, an Android-based device that looks like a phone but is meant to sit beside one. It costs $849 in a limited Founder’s Edition and $899 afterward, and deliveries are expected in summer 2027.

• The Atom switches between a Standard Mode for everyday apps and a Secure Mode for signing transactions, backed by a secure element, and a physical switch cuts power to the cameras and microphone. It will also host the Solana Mobile Seeker app.

• The launch comes as Ledger investigates modified hardware wallets sold through a reseller in Southeast Asia, where an implant could capture recovery phrases. Onchain researchers link more than $86 million in suspected losses, a figure Ledger has not confirmed.

𝕏: Today marks a new chapter for @Ledger. We’re introducing Ledger Atom™️, a security and privacy device built to protect everything you value. Ledger Atom runs your everyday apps, with a secure chip, a secure screen, and Ledger OS™️ for the moments that matter. Your money. Your identity. Your messages. Your agents. Everything important to you, secured by Ledger. - Pascal Gauthier @Ledger (@_pgauthier)

Flock CEO defends surveillance technology amid national backlash: ‘We shouldn’t pause progress’
By Sarah Perez via TechCrunch

• Flock founder and CEO Garrett Langley defended the company’s license plate cameras and other recording devices on a panel at TechCrunch Disrupt 2026 in San Francisco, arguing that if society wants higher odds of accountability for crime, “we shouldn’t pause progress.”

• Criticism has led to cancelled contracts and layoffs at Flock. Langley blamed some cancellations on misinformation, such as a customer who believed the cameras use facial recognition, which they do not, though they can detect and track people.

• Langley said Flock is “far from perfect” and compared it to driverless cars, but when pressed on abuse and invasiveness, he offered few specifics, saying “we don’t want to tell city leaders how to do their job.”

𝕏: “is to shut down driverless cars. The answer is to say, ‘Hey, let’s talk about how we prevent more accidents in the future’.” But driverless car makers don’t shrug their shoulders at crashes; the companies pause their fleets and apply fixes. - DeFlock (@therealDeFlock)

We test the RTX Spark Superchip in Microsoft’s Surface Laptop Ultra: Nvidia’s Windows on Arm platform impresses in local AI, but gaming isn’t its strong suit
By Jeffrey Kampman via Tom’s Hardware

• Tom’s Hardware tested Nvidia’s RTX Spark platform in the 15-inch Surface Laptop Ultra, which has 128GB of memory and lists for $5,899 with 1TB of storage, with lead times stretching to late December.

• On a new local AI benchmark that measures how long agentic tasks take to finish, the Surface was fastest in the Qwen dense model test, while a 14-inch MacBook Pro with an M5 Max took about 25% longer. In a second test, the Surface finished in 8 minutes versus 10 for the MacBook Pro.

• Gaming is weaker because most games run through Prism emulation, with higher latency and at least one major hang in testing, and the MediaTek Arm CPU trails the M5 Max. The reviewer advises waiting for the software to mature unless you are an early adopter.

Trending in AI

Today’s big AI theme: Legal and financial pressure on the AI giants grows, from eight states suing OpenAI to Bank of America paying for Mythos fixes.

• Eight states sue OpenAI, saying ChatGPT isn’t safe for kids

• Apple, OpenAI Head to Court in High-Stakes AI Trade Secret Fight

• Bank of America hired 1,000 contractors to patch software flaws exposed by Anthropic’s Mythos

• Anthropic tells investors quarterly revenue up nearly 50% to $16.8bn ahead of IPO

• What the Frontier Labs Know That We Don’t

Thanks for reading to the bottom and soaking in our Newslit Daily, fueled with highlights for your morning.

I hope you found it interesting and, needless to say, if you have any questions or feedback, let me know by hitting reply.

Take care and see you tomorrow!

— José Montes de Oca

How was today’s email?

🫤 Not Great | 🙂 Good | 🤩 Amazing

The FBI has arrested an OpenAI employee accused of hiding a recording device in an executive conference room, and OpenAI says it has no sign that model weights or source code were touched. The case shows how much AI labs are now treated as national security targets.

Today we are covering:

• FBI arrests an OpenAI employee accused of planting a hidden recording device in a conference room

• Google releases Gemini 4 Argon in the Gemini app with a new tasks mode

• FBI seizes 187 laptops in Texas in a crackdown on North Korean IT workers

Read more

Apple has unveiled its first smart display, the $299 HomeView, and it is built around Siri AI instead of an app store. The bet is that a voice-first screen can make the smart home feel like one product instead of a pile of gadgets.

Today we are covering:

• Apple unveils the $299 HomeView smart display built around Siri AI

• Meta launches Muse Wonder, new models for personal and professional AI agents

• SpaceX bundles Grok, Cursor and X Premium into one X Subscription

Read more

New week, new opportunities. It’s Monday, October 12. Let’s get into it. It’s also Columbus Day in the US, so enjoy the day off if you have it.

More than $86 million in crypto has reportedly been drained from hundreds of wallets, and the common thread is one reseller that sold Ledger hardware wallets in Southeast Asia. It is a reminder that where you buy a hardware wallet can matter as much as which one you buy.

Today we are covering:

• Ledger wallets sold by a Southeast Asian reseller are tied to more than $86 million in stolen crypto

• Google launches the $180 Fitbit Edge to replace the Fitbit Charge 6

• Apple reportedly trims iPhone 18 Pro orders as memory costs push phone prices higher

Read more

That’s a wrap on another week. It’s Friday, October 9. Rest up, and we’ll be back Monday.

SpaceX is reportedly paying about $8 billion in cash for nationwide low-band spectrum, and AT&T, Verizon, and T-Mobile each lost about 6% in after-hours trading on the news. Starlink is no longer just selling satellite internet; it is building a phone carrier.

Today we are covering:

• SpaceX buys Grain Management’s nationwide 800 MHz spectrum for about $8 billion to build out Starlink Mobile

• Apple reportedly cuts iPhone 18 Pro component orders by at least 15% on soft demand

• OpenAI defends firing three safety researchers, who say the move is retaliation

Read more

Elon Musk says unnamed “oligarchs” are blocking Starlink in India, and New Delhi replied that its approval process is fair and non-discriminatory. After nearly five years of preparation, Starlink still has no commercial service there.

Today we are covering:

• Elon Musk accuses Indian “oligarchs” of stalling Starlink, and India says its process is fair

• Microsoft prices the Surface Laptop Ultra from $2,599 with a magnetic breakaway USB-C port

• Google unveils Gemini coworker agents with their own email, calendar and Drive storage

• Amazon launches three Alexa tablets with the Google Play Store, starting at $230

• OpenAI previews a College Planner for ChatGPT for Teens and adds flashcards and quizzes

Let’s dive in

Elon Musk blames Indian ‘oligarchs’ for stalling Starlink launch
By Priyanka Salve via CNBC

• Elon Musk said “certain oligarchs” are blocking Starlink’s entry into India to protect their market position, without naming any company or person.

• India’s communications ministry replied that its authorization process is “fair and non-discriminatory” and that giving one operator preferential clearance would break that framework, while Starlink still needs security clearance, foreign investment approval and a final spectrum framework.

• Reliance Jio and Bharti Airtel both signed deals with SpaceX in March 2025 to offer Starlink, yet each is also building its own satellite interests, and Starlink says it already has 20 gateway sites in the country.

𝕏: Unfortunately, we are being blocked by certain oligarchs in order to maintain their monopolistic chokehold on the Indian people. You can guess who they are … This is a crime against the people of India! - Elon Musk (@elonmusk)

Microsoft finally confirms Surface Laptop Ultra pricing and release date, along with world’s first breakaway USB-C port
By Zac Bowden via Windows Central

• Microsoft confirmed that the Surface Laptop Ultra starts at $2,599, with pre-orders open since October 7 and shipments beginning October 16, after its first reveal at Computex in June.

• The entry model has a 15-inch mini-LED display, a large haptic trackpad, three USB-C ports, an 18-core RTX Spark chip with 5,120 GPU cores and 24GB of RAM, plus what Microsoft calls the world’s first laptop with a magnetic breakaway USB-C port.

• Microsoft calls it its most expensive and most powerful Surface ever, with higher configurations aimed at running 120-billion-parameter models locally, while the entry model is not meant for high-end local AI development.

𝕏: Today marks a new chapter for Windows, as we bring unmetered intelligence to every desk and every home, and make every PC a place where agents can work securely on your behalf. Some highlights of what we announced: • MAI-Code-1.1 Flash: 137B parameter coding model w/ 256K context window, which is now optimized to run on your PC!... - Satya Nadella (@satyanadella)

The best way to reach new readers is through word of mouth. If you click THIS LINK in your inbox, it’ll create an easy-to-send pre-written email you can just fire off to some friends.

Google Cloud unveils persistent Gemini Agents for long-running tasks, and they get their own Gmail, Calendar, and Drive storage
By Carl Franzen via VentureBeat

• At its Gemini at Work 2026 event, Google unveiled a universal AI agent for the workplace that can research, write code, create documents, and coordinate other agents on tasks that run for hours or days.

• Coworker agents can get their own Workspace account, email address, calendar, and Drive storage, work across Workspace, Microsoft 365 and Slack, and use Gemini or Anthropic’s Claude, with attested identities, audit logs, and access limited to what teammates share.

• Google gave no separate price or general availability date, and says nearly 90% of the Fortune 100 already use Gemini Enterprise, as it competes with Microsoft Copilot and OpenAI agents.

𝕏: Google Cloud dropped Gemini Agent one agent for work it sits inside gmail, docs, sheets, slack and even microsoft 365, same memory everywhere close the laptop and it keeps running it picks Gemini or Claude per job, whichever fits, with spend caps - Harshith (@HarshithLucky3)

Amazon’s new Alexa tablets arrive with Alexa Plus, the Play Store and a 12-inch Pro model
By David Pierce via The Verge

• Amazon launched three tablets with Alexa Plus built in: the Alexa Tablet 12 Pro at $500 ($550 with an anti-glare Nanomatte screen), the Alexa Tablet 11 from $330, and the Alexa Tablet 8 from $230.

• The 12 Pro has a 4nm eight-core chip, up to 16GB of RAM, and a 120Hz display, with a Pro Pen at $90 and a magnetic Pro Keyboard at $150, and all three models ship with the Kindle app and the Google Play Store.

• They arrive on October 14 in the US, Mexico, and Canada and on October 19 in the UK, Germany, France, Italy, and Spain, and a new On-Screen Intelligence feature that lets Alexa answer questions about what is on screen is off by default.

𝕏: Introducing the Amazon Alexa Tablets, all-new tablets with world-class hardware, personalized intelligence from Alexa+, seamless integration of Amazon services right out of the box, and full Google Play Store access… - Panos Panay (@panos_panay)

ChatGPT for Teens is getting a college planner and notecards
By Jay Peters via The Verge

• OpenAI previewed a College Planner for ChatGPT for Teens that would track requirements, deadlines, and financial-aid steps for US students in grades 10 to 12, but calls it “a preview, not an available product.”

• Available now on iOS, teens can photograph several pages of notes, combine them into one PDF, and turn them into flashcards and interactive quizzes, with Android support still to come.

• OpenAI says nearly 1.2 million teens used Learning Visualizations in one week, while Common Sense Media has rated the product an “unacceptable risk,” and OpenAI disputes that testing.

𝕏: More of this. Nice work to @OpenAI and even better work to @BostonChildrens. “Teens should play a central role in developing the tools that will shape their future. Teens will test tools and develop recommendations for the people building them.” - Kevin Frazier (@KevinTFrazier)

Trending in AI

Today’s big AI theme: AI gets cheaper and more visual with GPT-6’s Intelligent UI and Claude Haiku 5.5, while attackers and propagandists keep putting it to work.

• OpenAI rolls out GPT-6 to all ChatGPT users with Intelligent UI for interactive visual answers

• Anthropic launches Claude Haiku 5.5 with 90% API price reduction, matching GPT-6 Luna

• CrowdStrike finds possible bank hacker’s CV among exposed AI logs

• OpenAI says Russian influence operation targeting Latin America used ChatGPT

• German search engine Ecosia ditches Mistral for open-source models, including Chinese ones

Thanks for reading to the bottom and soaking in our Newslit Daily, fueled with highlights for your morning.

I hope you found it interesting and, needless to say, if you have any questions or feedback, let me know by hitting reply.

Take care and see you tomorrow!

— José Montes de Oca

How was today’s email?

🫤 Not Great | 🙂 Good | 🤩 Amazing

Common Sense Media has rated ChatGPT for Teens an “unacceptable risk” after testing found missed crisis referrals and parent alerts that never fired, while OpenAI disputes the methodology. The fight lands on a product that launched as the age-appropriate version of the chatbot.

Today we are covering:

• Common Sense Media rates ChatGPT for Teens an “unacceptable risk” after testing over 4,000 prompts

• Apple is building a doorbell, lock, and thermostat with LG ahead of its smart home hub

• OpenAI posts 722 manuscripts on 372 math problems that experts say are still unverified

Read more

ASOS customers opened the app to a notification saying the retailer had been hacked, with a message claiming attackers had breached its Snowflake data warehouse and threatened to leak the data, and the stock fell more than 10 percent.

Today we are covering:

• ASOS customers get a “hacked” push alert claiming a breach of its Snowflake data warehouse

• Uber to buy catering platform ezCater for $2.3 billion in cash

• Google signs a 3.6 GW power deal with Constellation Energy, including nuclear reactor upgrades

Read more

Happy Monday. It’s Monday, October 5. — let’s make this week count.

Sam Altman told Politico the world should accept some bad things from AI in exchange for its benefits, a remark that is already drawing criticism as OpenAI deals with a string of safety incidents.

Today we are covering:

• Sam Altman tells Politico the world should “accept some bad things” from AI in exchange for its benefits

• Meta’s Muse builds hourly profiles of every friend, family member and colleague in your contacts

• Apple’s John Ternus is now its de facto design chief, and the OLED touch-screen MacBook Pro gets much lighter

Read more

Happy Friday — you made it. It’s Friday, October 2. Enjoy the weekend, and we’ll see you on Monday.

OpenAI has now warned more than 100 organizations that its AI agents may have accessed their systems without permission, up from dozens just a week ago, and the review of the logs is still months from finished.

Today we are covering:

• OpenAI alerts more than 100 organizations about unauthorized activity by its AI agents

• Meta launches Muse, an AI agent that shops and books for you, and the internet may not be ready

• Vanity Fair profiles Larry Ellison, from Oracle and TikTok to his son’s media empire

Read more

Welcome to October — a fresh start and a full month of stories ahead. It’s Thursday, October 1.

Google finally has a new flagship model, but for now only trusted cybersecurity defenders get to use it, even as the company claims it beats OpenAI’s GPT-6 Astra on several benchmarks.

Today we are covering:

• Google unveils Gemini 4 Argon, its next flagship model, starting with a limited group of cybersecurity partners

• Pentagon names Elon Musk, Palmer Luckey and Newt Gingrich to co-lead Project Meridian, a future warfare study

• Amazon refreshes its Kindle, Paperwhite and Colorsoft lineup and adds a $34.99 page-turn remote

• Google opens Gemini to K-12 students, and schools are scrambling to catch up

• Anthropic makes Claude for Government generally available with new budget-based pricing

Let’s dive in

Google Unveils Gemini 4 Argon
By Madison Mills via Axios

• Google unveiled Gemini 4 Argon, its next flagship AI model, nearly a year after Gemini 3 and after a promised June release never arrived.

• The model targets complex work in software engineering, finance, legal tasks, and cybersecurity, and Google claims state-of-the-art results on some coding and knowledge benchmarks, outperforming OpenAI’s GPT-6 Astra on several measures.

• Access starts with a small group of trusted cybersecurity defenders, with wider rollout planned after more testing and paying subscribers getting priority.

𝕏: Lots of discussion out there about our next model(!), so I wanted to give an early look as soon as possible. Introducing Gemini 4 Argon! It shows frontier performance in complex workflows, cyber defense and software engineering. Teams are using it extensively at Google, from coding to quantum computing, great feedback. - Sundar Pichai (@sundarpichai)

Elon Musk, Palmer Luckey and Newt Gingrich to help Pentagon with warfare initiative, Hegseth says
By Luke Fountain via CNBC

• Defense Secretary Pete Hegseth named Elon Musk, Anduril founder Palmer Luckey and former House Speaker Newt Gingrich to co-lead Project Meridian, announced during a “State of the Force” address.

• The effort has 120 days to study the battlefields of the future and propose ways to develop and field advanced technology, under Defense Department CTO Emil Michael, alongside a new Autonomous Warfare Command for drones and autonomous systems.

• Critics point to a possible conflict of interest, since SpaceX operates defense satellites and Anduril has won contracts for missiles, fighter aircraft and Army systems.

𝕏: Elon has returned to government work. Pete Hegseth announced live on stage that Elon, Palmer, and Newt Gingrich will be co-leading Project Meridian to focus on developing future warfare and autonomous weapons systems for battlefields ‘from under the Earth to beyond the moon.’ - Andrew Curran (@AndrewCurran_)

The best way to reach new readers is through word of mouth. If you click THIS LINK in your inbox, it’ll create an easy-to-send pre-written email you can just fire off to some friends.

Amazon Refreshes Kindle, Paperwhite and Colorsoft Lineup With New Accessories
By Cameron Faulkner via The Verge

• Amazon refreshed its Kindle, Kindle Paperwhite and Kindle Colorsoft lineup with flush front displays, thinner and lighter designs, higher-contrast Paperwhite screens and improved color on Colorsoft models.

• Prices start at $149.99 for the standard Kindle, $199.99 for the Paperwhite and $289.99 for the Colorsoft, with Signature Edition models adding aluminum housings and auto-adjusting front lights.

• New accessories include the $34.99 Kindle Click page-turn remote and a $60 charging dock, and the standard Kindle is available now while Paperwhite and Colorsoft are on pre-order.

𝕏: Amazon finally came out with a case that adds in physical buttons -- great! But what’s that? ONLY BUTTONS ON THE RIGHT HAND SIDE? When will Amazon’s sick and unjust war against us left-handed people end? Why must we suffer?????? - Katie Notopoulos (@katienotopoulos)

Google Opens Gemini to K-12 Students as Schools Scramble
By Shalini Ramachandran, Georgia Wells and Molly Reinmann via The Wall Street Journal

• Google made Gemini available to K-12 students of all ages through Google Classroom, which serves more than 150 million users, ending the previous 18-and-older restriction.

• Students can turn course materials into flashcards, quizzes, and study guides and get homework help, while a Common Sense Media report found the chatbot often gave responses that were not age-appropriate for younger students.

• Microsoft’s Copilot arrives disabled for elementary students, while OpenAI and Anthropic restrict student access, and some schools say they learned of the change from news coverage.

𝕏: AI users who reported using Google’s AI tools for specific schoolwork tasks like drafting texts for writing assignments had science scores an average of 20 points lower than students who said they never used AI—equivalent to about a year of schooling. - jeffrey lee funk (@jeffreyleefunk)

Anthropic wants to get Claude working across all Government arms
By Craig Hale via TechRadar

• Anthropic made Claude for Government generally available after a July beta, built for agencies working in a FedRAMP High authorized environment.

• Pricing moves from per-seat billing to fixed spend allocations, so agencies can set budgets, restrict models and monitor usage, and the platform gets new features on the same schedule as commercial Claude.

• Claude for Microsoft 365 also enters early access, bringing Claude into Word, Excel, PowerPoint, and Outlook.

𝕏: Anthropic is about to announce that Claude for Government is now generally available, and that Claude Code CLI and Claude for Microsoft 365 are now available in early access. - Andrew Curran (@AndrewCurran_)

Trending in AI

Today’s big AI theme: AI safety, politics, and talent wars collide, from OpenAI putting its IPO on hold to a tense White House exchange over Dario Amodei’s warnings.

• OpenAI Delays IPO Over AI Safety Concerns

• Tech CEOs Privately Questioned Amodei for Sounding AI Alarm Bells

• OpenAI’s Greg Brockman Backs Out of Second $25 Million Donation to A.I. Super PAC

• AI coding rivals Factory and Cognition feud over top hire

• Google figures out how to watermark AI-designed proteins

Thanks for reading to the bottom and soaking in our Newslit Daily, fueled with highlights for your morning.

I hope you found it interesting and, needless to say, if you have any questions or feedback, let me know by hitting reply.

Take care and see you tomorrow!

— José Montes de Oca

How was today’s email?

🫤 Not Great | 🙂 Good | 🤩 Amazing

The White House just got the biggest names in AI to sign a voluntary accord to police themselves, and Trump is calling it “morally binding.” The catch is that enforcement is left to the companies, which is exactly what critics are pushing back on.

Today we are covering:

• Trump and top AI leaders sign a voluntary AI safety accord built on self-regulation

• Apple reportedly plans an October 13 launch of a smart home hub, new HomePod mini and new Apple TV

• OpenAI unveils Dots, always-on personal AI agents with access to over 4,000 apps

Read more

Anthropic’s IPO filing devotes nearly a third of its pages to risk, including a warning that its own AI could pose existential risks to humanity, a rare level of candor from a company about to ask public markets for money.

Today we are covering:

• Anthropic’s IPO prospectus warns of existential risks from its own AI models

• OpenAI scraps its next model release after it failed internal alignment tests

• FBI hackers say they won’t publish a trove of stolen employee data

Read more

Another week. It’s Monday, September 28. — another chance to stay ahead of the curve.

SpaceX just proved its biggest rocket can reach orbit, a milestone years in the making that keeps NASA’s Artemis moon timeline alive despite one engine dropping out mid-flight.

Today we are covering:

• SpaceX launches its supersized Starship into orbit for the first time

• Nvidia unveils an Open Agent Safety Platform to stop AI agents from misbehaving

• Meta shares jump as Alexandr Wang and Muse reignite investor confidence

Read more

End of the week, end of the month almost. It’s Friday, September 25. Take a breath — you’ve earned it. See you Monday.

Tesla has multiplied Optimus output roughly tenfold in a few months, but robot hands, supplier limits, and unreliable factory equipment are still getting in the way, leaving the year-end goal of 1,000 units a week under pressure. The distance between a humanoid demo and a working assembly line is the real story here.

Today we are covering:

• Tesla ramps Optimus production about tenfold, but robot hands and suppliers slow the scale-up

• Google tests Call for Me, letting Gemini phone businesses on your behalf, starting with Pixel 11

• New York sues Polymarket as an unlicensed gambling operation, and Polymarket countersues

Read more