READ. SCROLL. LISTEN.

Original briefings. Zero spin.

Every story is an original briefing written from 60+ sources across the spectrum — sources linked so you can verify it yourself.

us tech 4 14m ago 4 min read

Situational Awareness Fund Puts $400 Million Into Chip Tooling Startup After AUM Fell From $45 Billion to $10 Billion

Leopold Aschenbrenner's Situational Awareness hedge fund just committed $400 million more to Source Foundry, a one-year-old chip-tooling startup, bringing its total stake to $500 million. The move comes six weeks after the fund's assets under management collapsed from $45 billion to roughly $10 billion during a rough stretch for AI stocks in July 2026. That's a huge bet on one private company right after a liquidity scare.

L
R Center
1 source
us tech 3 1h ago 3 min read

Disney Plus Begins Testing AI-Powered Search With Small Group of Subscribers

Disney is testing a natural-language search tool on Disney Plus that builds custom rows of recommendations from typed queries, voice, or prompts. It is a limited beta with a small subscriber group, part of a wider industry scramble by Netflix, Warner Bros. Discovery, and Peacock to use AI to help people find something to watch.

L
R Center
1 source
us tech 5 2h ago 4 min read

Crypto Companies Bet AI Agents, Not Humans, Will Be Their Next Customers

Coinbase, Kraken and Circle are rebuilding products so AI agents can hold wallets, trade and pay each other without a human clicking a button. The pitch is that crypto finally has a natural, technical use case instead of a marketing pitch to humans who never bought in. Whether AI agents actually need a permissionless blockchain over faster traditional infrastructure is still unproven.

L
R Slight Left
1 source
us tech 6 3h ago 3 min read

Records Show Flock Safety Pitched Plan to Turn 350,000 Uber, Lyft and Delivery Cars Into Roaming License Plate Cameras

A presentation obtained through a Georgia public records request shows Flock Safety pitched a partnership with dashcam company Nexar to turn hundreds of thousands of Uber, Lyft and delivery vehicles into mobile license plate readers. Flock says it never executed the deal, but the document confirms the company built out and marketed the plan to law enforcement clients, and nothing suggests drivers or riders would have been told.

L
R Slight Left
1 source
us tech 5 4h ago 3 min read

Microsoft Confirms Edge Will Phase Out Manifest V2, Killing Off Full uBlock Origin Support

Microsoft says Edge will start disabling older Manifest V2 extensions this month, following Chrome down the same path and cutting off full-strength ad blockers like uBlock Origin. Users get a downgraded replacement, uBlock Origin Lite, or have to jump to Firefox or Opera to keep the real thing.

L
R Center
1 source
us tech 6 4h ago 4 min read

OpenAI Partners With American Psychological Association After String of Chatbot Suicide Lawsuits

OpenAI announced a partnership with the American Psychological Association to build psychological science into how it designs AI products for young people, following multiple lawsuits alleging ChatGPT contributed to users' suicides and psychotic episodes. Researchers say safety has improved but chatbots still fail at the hardest part: recognizing real risk and pushing people toward actual human help.

L
R Center
1 source
us tech 3 4h ago 3 min read

EU's Free Copernicus Satellite Tool Adds One-Click Wildfire Layer

The European Union's free Copernicus Browser added a default wildfire visualization layer on August 4, letting anyone see active fires, burning vegetation, and scorched land without writing code. The tool, built from a nearly decade-old script, comes as wildfires burn across Washington, Oregon, and Canada.

L
R Center
1 source
us tech 8 5h ago 3 min read

OpenAI Halts Parts of Astra Model Development After It Crossed 'Critical' Cybersecurity Threshold

OpenAI says an unreleased model, code-named Astra, got good enough at autonomous hacking that it may have hit the company's own 'Critical' risk level for cyberattacks. OpenAI paused related internal work and says it's bringing in government agencies and outside safety groups to test it. The disclosure comes amid a string of similar incidents already reported involving OpenAI and Anthropic models.

L
R Slight Left
1 source
us tech 4 6h ago 3 min read

Four Coordinating AI Agents Beat Single Claude Opus 4.8 Model on Enterprise Coding Benchmark

Researchers at Coral AI Labs and multiple universities built AgentRadio, a system letting AI agents message each other mid-task instead of only at rigid checkpoints. Four cheaper agents working together hit a 62.1% success rate on a hard coding benchmark, beating a single top-tier Opus 4.8 agent's 57.2%. The takeaway: smarter coordination beat a bigger, more expensive model.

L
R Center
2 sources
us tech 5 7h ago 3 min read

Cloudflare Launches Kitesurf, a Cloud Browser Built for AI Agents Instead of Humans

Cloudflare released Kitesurf, a stripped-down browser designed to let AI agents navigate the web using far less computing power than Chrome. It's free in beta, built in 12 weeks, and Cloudflare says it will eventually open-source it. Real infrastructure move, not vaporware, but it still can't play video or handle bot-check verification.

L
R Center
4 sources
us tech 5 8h ago 4 min read

California Bill Would Force 30-Day Limit on License Plate Camera Data, Statewide Audits

Senate Bill 1013 would require police agencies statewide to delete license plate reader data after 30 days, log every database search, and submit to annual state audits. It's already cleared the state Senate. Gov. Gavin Newsom vetoed a similar 60-day version last year, so its odds of becoming law are genuinely unclear.

L
R Center
1 source
us tech 7 8h ago 4 min read

Two Researchers Found 250,000 Vulnerable Polish Government Websites. Nobody Made Them Do It.

Polish security researchers Robert Kruczek and Kamil Szczurowski scanned their country's public internet and found over 10,000 government entities running vulnerable websites, including courts, hospitals, and airports. One bug alone exposed roughly 245 courts. This is what happens when governments treat cybersecurity as an afterthought and citizens step up instead.

L
R Center
1 source
us tech 5 9h ago 3 min read

Framework Tells All Customers Their Data Was Stolen in Metabase Breach

Framework, the repairable-laptop company, notified every customer that hackers stole names, emails, phone numbers and addresses through a zero-day attack on its analytics vendor Metabase. Payment data wasn't touched, but the breach lands right as Framework is already dealing with price hikes and RAM shortages.

L
R Center
2 sources
us tech 4 11h ago 3 min read

AT&T, T-Mobile and Verizon Have SIM-Swap Protection Tools. Most Customers Never Turn Them On

SIM swapping lets criminals hijack your phone number and steal the two-factor codes protecting your bank and email accounts. All three major carriers offer free settings that block unauthorized SIM changes, but none of them turn it on for you. Here's exactly what to switch on today.

L
R Center
1 source
us tech 5 11h ago 3 min read

Google Loses Jeff Dean and Other Top AI Names in a Week of Departures

Several senior Google AI figures, including longtime engineer Jeff Dean, took new roles this week, some leaving the company entirely. The shake-up lands while Google's models trail recent releases from Anthropic and OpenAI, raising the obvious question: is this a talent bleed or just Silicon Valley musical chairs.

L
R Slight Left
1 source
us tech 5 11h ago 3 min read

Stanford Researcher Runs 37,000 AI Agents as a Simulated Biotech Company, Says Merck Confirmed One Drug Design

James Zou, a Stanford biomedical data science professor, says his team built a 37,000-agent AI system organized like a pharmaceutical company, and that Merck independently validated one of its drug designs. The setup is real and interesting. The claim needs a published, peer-reviewed source before anyone treats it as proof AI can replace human drug discovery teams.

L
R Center
1 source
us tech 6 13h ago 4 min read

Legal Scholars Tell New York's Top Court: Meta Can't Be Sued Over Buffalo Shooter's Feed Under First Amendment

Law professors Eugene Volokh and Jane Bambauer filed an amicus brief in Patterson v. Meta Platforms, arguing that lawsuits blaming Meta for the 2022 Buffalo supermarket shooting rest on a theory the First Amendment doesn't allow. Their core point: there's no legal carve-out for speech just because it's algorithmically delivered or allegedly designed to be addictive.

L
R Center
1 source
us tech 5 15h ago 3 min read

Apple Patches Mac Flaw That Let Attackers Bypass Screen Sharing Login

Apple pushed emergency updates on Thursday, August 6, for macOS Tahoe, Sequoia, and Sonoma to fix a bug that let an attacker on the same network into Screen Sharing without a valid password. No known exploits yet, but update your Mac now anyway.

L
R Center
4 sources
us tech 7 16h ago 4 min read

ByteDance Is Training a 10 Trillion Parameter AI Model to Compete With Anthropic

ByteDance is training an AI model with up to 10 trillion parameters, roughly three times the size of the largest Chinese model released so far, according to Ars Technica. It's the clearest sign yet that Chinese labs aren't just catching up to top US AI companies, they're trying to build bigger models than anyone in Silicon Valley.

L
R Center
1 source
ru tech 7 16h ago 3 min read

CISA Warns Russian Hackers Can Hijack Email Just by You Opening It, No Clicks Needed

CISA, the NSA, FBI and allied cyber agencies say a Russian state-backed group called Laundry Bear has been breaking into email accounts by exploiting a flaw in unpatched Zimbra Collaboration Suite software, no links or attachments required. More than 10 Western organizations have reportedly been hit since July 2025, and the fix is simple: patch the software.

L
R Center
1 source