UPDATES
Hey! Hope you all are doing well!
—
My response to an argument that most people don’t want to do things. This is the whole point of H3! ACTIVATION! It’s not people’s fault that we’ve been programmed to be passive and silent. We can’t just give up on people. We can try to revive them if they don’t feel a spark.
—
Attacker vs. Defender AI Advantage
I think the biggest thing that’s going to give attackers the advantage over defenders in this AI race is the fact that defenders are embedded in bureaucracy.
Attackers can move from an idea to implementation within hours
Most companies still need days, weeks, or months
Bureaucracy creates the biggest defensive disadvantage
Change control adds friction, even inside effective organizations
This is an organizational problem before it’s a technology problem
Read the blog:
—
I made a site so I can study the Te Ching (about time)
—
A few other new blogs:
A new blog on personified agents.
—
My early thoughts on Jev and how I’m using it in LifeOS.
—
If you’ve been messing with Jev this is a must-read.
Sponsor
The IT and security field guide to AI adoption
AI is everywhere right now. But for many teams, the reality hasn’t matched the promise.
Tools that look great in demos don’t hold up in real workflows. And instead of reducing workload, AI can introduce new risks and oversight.
So what’s actually working?
Tines just released a guide that takes a more practical look at AI adoption. Inside, you’ll get:
-A practical framework for evaluating tools beyond the demo
-A step-by-step approach to selecting tools that hold up in production
-Key questions to ask before committing to a vendor
-Best practices for keeping humans in the loop
If you’re thinking about AI beyond experimentation, this is a useful place to start.
CYBERSECURITY
AI Security Is an Engineering Problem My friend Saša’s post as the new CSO of NVIDIA lays out controls for agents, including restricted credentials, sandboxed tools, network limits, protected logs, and human approval for consequential actions.
Boundaries must survive bad decisions
Agents need traceable identities and credentials limited to their assigned tasks only
Human approval still gates consequential actions
OpenShell sandboxes agents while policies govern files, networks, and system resources at runtime
Failed tests become repeatable checks for future releases
I think this is a great message and a big part of what we need to do to prepare for what’s coming. NVIDIA ARTICLE
Claude Opus 5 Helped Researchers Take Over OpenAI Staff Accounts via Chained Bugs Researchers used Claude Opus 5 to chain a forum image bug with OpenAI's login flaw and reach staff accounts. OpenAI fixed the flaw within 14 hours and paid $6,500. THE HACKER NEWS ARTICLE
AI is set to help cyber attackers much more than defenders An argument that AI attacks are easier to automate because success is obvious, while defensive mistakes can break live systems. Attackers get a clean signal. THE RECORD ARTICLE
Anthropic-linked CVEs pile up, attackers mostly shrug VulnCheck tracks 225 vulnerabilities tied to Glasswing and Anthropic, but only one, a critical Ghost SQL injection, has been exploited in the wild, according to this. I’m not sure how they’re so sure what all’s been exploited in the wild but the analysis matches what Jeremiah Grossman has been talking for a while about regarding there being a major difference between AI finding lots of bugs and them being used in real attacks. THE REGISTER ARTICLE
Flock cameras are tracking people as well as cars MALWAREBYTES ARTICLE
Flock cameras detect people too
One recovered camera held thousands of vehicle detections despite Flock's encryption claims
More than 2,000 agencies accessed Alpharetta records
Washington MPD used Flock data to track officers under investigation without their knowledge
Safeguards need deletion, audits, and search limits for ordinary people
NATIONAL SECURITY
Autonomous NATO strike drone uses Nvidia Jetson Orin Nano Scaleout Systems flew a low-cost loitering munition that found, ranked, and bombed an armored vehicle. And onboard AI completed the mission in under 320 seconds without communications. TOM'S HARDWARE ARTICLE
Russia plans new navigation system for Shaheds, guided bombs Russia is considering Ariadna for Shahed drones and KAB guided bombs, giving them ground-based navigation when satellite links fail. The system reads signal delays from three fixed beacons and could guide weapons up to 100 kilometers into Ukraine. Ukraine uses similar tech on reconnaissance drones. NV ARTICLE
AI
Instinct is the new agent hotness Lots of people talking about how good this one is. I tried it, but didn't give it any data, so it didn't really do much for me because I already have Kai working inside of LifeOS, but I've seen extremely impressive results from other people.
They raised $250 million at a $2.5 billion valuation after going viral with invite-only AI assistant access. Now it faces Meta's Muse. The founder, Shinn, is 23. BUSINESS INSIDER ARTICLE | GET INSTINCT
Study Finds Enterprises Rethink Work to Get More From AI Deep adopters reported AI returns 93% of the time, versus 55% for shallow adopters. The difference is how well the org is prepared for articulating how it works to AI. PYMNTS ARTICLE
What AI Anxiety Looks Like in China Chinese workers fear AI will take their jobs, but the government still wants AI everywhere by 2035. THE ATLANTIC ARTICLE
How Boko Haram Fighters Are Using AI To Make Explosives, Plan Attacks Boko Haram fighters used AI chatbots for battlefield advice. Former members described dedicated units, jailbreaks, and help with explosives and drones. TIMES OF INDIA ARTICLE
TECHNOLOGY
China Holds 90% of World's Commercial Robot Data China opened 22 humanoid robot centers in 14 months after its 2023 policy directive. Scale AI says production costs run 60% below America's for training data. SEOUL ECONOMIC DAILY ARTICLE
If AI coding is lowering your code quality, you’re not managing quality AI coding agents can raise output 2–3x while bugs stay flat or fall, but only when requirements, tests, manual checks, reviews, and monitoring work together. IOURI KHRAMTSOV POST
The Senior Engineer Death Spiral Senior engineers can vanish into ambitious projects, and Sunil Pai turns that failure pattern into a practical reset: Basically: share work, help teammates, rebuild momentum, then earn bigger projects. SUNIL PAI POST
Korea raises data breach fines to 10% of revenue Korea's regulator can fine companies up to 10 percent of revenue for massive negligent leaks, and firms get reductions for prevention and 72-hour warnings. KOREA JOONGANG DAILY ARTICLE
HUMANS
The art of letting someone be mad at you Dr. Nicole LePera argues that people pleasers should stop rushing to erase someone’s anger and instead reflect, ask what they need, and respect their timeline. She says it includes tolerating temporary disapproval without turning guilt into self-punishment. DR NICOLE LEPERA THREAD
Unemployment: A National Misunderstanding Worth reading: Jim Clifton says the standard 4% unemployment rate misses discouraged workers, part-time workers, and prime working-age adults, putting real unemployment near 12%. JIM CLIFTON ARTICLE
Gallup counts 6.5 million unemployed
U-6 adds discouraged workers and part-time workers seeking full-time jobs too
Clifton counts 22 million out of work
A monthly survey of roughly 60,000 households produces the headline unemployment number for America
AI job losses could deepen this gap for millions
Write while learning Beginners often need the missing bridge between simple explanations and expert docs; purplesyringa says writers should record confusing steps while they still remember them. PURPLESYRINGA BLOG
Grandparents send AI storybooks to their unhappy children TECHRADAR ARTICLE
In an Age of AI Answers, Asking Why Matters More This conversation asks what universities still need to teach. SEOUL ECONOMIC DAILY ARTICLE
5 ways to become more “high agency” in the age of AI AI can make people capable yet passive. Agency grows when you choose and act. BIG THINK ARTICLE
IDEAS
Personal agents optimize outcomes, but people still want the process Personal agents should preserve the browsing and comparison people enjoy. Alex Kehr sees few recurring tasks worth delegating. Strong disagree but interesting opinion, and I think he’s right about certain parts. ALEX KEHR POST
Write while learning Beginners often need the missing bridge between simple explanations and expert docs; purplesyringa says writers should record confusing steps while they still remember them. I think this is really good advice, and it's a big part of how I've been blogging for all these years. PURPLESYRINGA BLOG
I am often wrong Product teams can solve hard problems by separating facts, problem definition, approach, goals, and action, then revising the plan when new information arrives. Boris Cherny treats being wrong as useful feedback, especially when a simple plan and clear success criteria are missing. Six steps, repeated many times daily. BORIS CHERNY BLOG
DISCOVERY
California’s wine country is drowning in unwanted grapes Wine sales fell 23% from 2020 to 2025, leaving half of California’s grape harvest without buyers and growers weighing losses, abandoned crops, or vineyard removal. FORTUNE ARTICLE
Pulsars changed the timing models scientists use for gravity BIG THINK ARTICLE
Are You Working? Creative work often looks like idleness, but Frederick’s mice discover that poems and mathematics gather light for winter when food and conversation run out. THE MARGINALIAN ARTICLE
Am I Cited by AI Scite lets researchers see which AI tools read their papers and cite them. Its index tracks 19.4 million AI reads. SCITE TOOL
How Hacker News ranking really works Ken Shirriff's blog tracks Hacker News penalties, showing popular stories can vanish despite strong upvote scores. KEN SHIRRIFF BLOG
What China’s Reading Five Chinese books follow women and men through lost friendship, fashion, family trauma, wartime memory, and academic power. Most slipped past major prize lists. CHINA BOOKS REVIEW ARTICLE
RECOMMENDATION OF THE WEEK
I highly recommend the UL Book of the Month, The Creative Act, by Rick Rubin.
Can’t wait for book club this weekend!
APHORISM OF THE WEEK
The more important something is, the more we have to struggle in the attempt to reduce it to language.
GET THE MEMBER EDITION
You’re currently receiving the STANDARD edition.
Members help this work continue. If you enjoy the newsletter, the podcast, what I put on YouTube, or any of my open-source projects on Github, I ask you to please become a member. It allows me to stay focused on learning and building and sharing. It’s like a cup of coffee or two per month.
Plus, members get numerous benefits, including:
25-50% off all UL Paid Content, including the upcoming Human 3.0 / AUGMENTED ONLINE portal!
Access to the extraordinary UL Member Community that includes vibrant conversations with ~1,500 of the smartest and kindest people you’ll find on the internet
Member-only Content, such as EDC guides on tech stacks, personal productivity routines, my recommendations on Critical skills to Build Going Forward, Trend Identification and Analysis, and more…
Access to the Member Archive of previous Member-only content, the Book Club archive, etc.
Access to The UL Book Club that’s been going monthly since 2017! One of the highlights of my and many attendees’ month!
Access to the Monthly Member Meet-up where we talk about our routines, productivity workflows, what’s on our minds, etc.
Access to In-Person Events like our dinners in Vegas, San Francisco, etc.
And much more coming…
This is the moment to connect with others who are smart, kind, and asking the same questions we are. Where is this all going? And how do to prepare?
Join the conversation.