Director of Marketing CBD American Shaman, SEO, Partial Hippie, Techie, News #Crypto Obsessed. CBD evangelist. Research nut on topics needing exposed.
Kansas City, MO
Joined October 2007
- Tweets57.5K
- Following46.4K
- Followers82.7K
- Likes15.8K
Joel Mackey retweeted
Someone had a computer play Monopoly 312 million times to find the best strategy. The winning bot won 43% of games. Its strategy: buy every property you land on, never hoard cash, and take out mortgages to get three houses as fast as possible. Pretty sure it works outside the board game too 👀
Joel Mackey retweeted
Claude Code just found a secret North Korean virus called "CRIMSON-9" taking over iPhones, iPads, and MacBooks, and nothing has ever spread this fast…
one person accepts a single Bluetooth file, and from then on it jumps to every Apple device nearby on its own, no tap, no download. It is already on 1.7 million of them.
I am not a security expert. I can only tell you what one AI agent found on two phones in my house last week.
My brother walked in from the other room, scared, holding a phone that was doing things no one asked it to. He had never touched it. I grabbed a second phone and filmed it, because I knew nobody would believe me.
By the time I looked up it was finished. His phone was dead, locked, mocking him on the screen, and there was nothing left to save.
Then it hit me that his phone and mine had been on the same network. So that night I ran Claude Code across every device I own.
On my phone it found a connection I never made, quietly sending data to an address it traced back to North Korea. It flagged the pattern as something it had never seen before and tagged it CRIMSON-9.
As far as it could tell, it started with one file my brother accepted over Bluetooth a week ago, something he thought was a photo. After that nobody had to tap anything. It moved by itself between devices that were simply near each other.
Here is what Claude Code showed me on our phones.
→ it tagged the thing CRIMSON-9, talking to a server it placed in North Korea
→ it got in through one Bluetooth file, then moved between nearby devices by itself
→ it had copied our photos, our logins, and the codes our banks text us
→ then it locked the phone and reached for the next device in range
If it can cross a room in one house in an evening, I do not want to picture a train car or an office.
I asked Claude Code how to stop it, and it went through every device I own and shut every way in.
Here is what I did.
→ turned Bluetooth off whenever I am out
→ turned off the setting that lets nearby devices connect without asking
→ set Claude Code to watch everything on my network and block any device it does not know
→ stopped my phone from trusting cables and computers it has never seen
Now nothing unknown can reach my phone, my laptop, or my kid's tablet without my say. Claude Code blocks every connection it did not approve.
My brother lost everything on his phone. I still do not know exactly what it was, but I know Claude Code caught mine before it got that far.
The full step by step and the exact Claude Code setup are in the first comment. It takes ten minutes. Save this, and send it to anyone you want to keep safe.
Joel Mackey retweeted
JUST IN: UC Berkeley math professor stops grading homework for correctness, saying widespread AI use has made it "pointless," and now awards credit simply for turning in assignments.
Joel Mackey retweeted
The length of software tasks AI agents can complete has doubled roughly every 7 months since 2019, and recent models are moving even faster, from 30 seconds of human work in 2022 to a full workday today.
Yet inside most companies, AI still isn't trusted to reliably update records in Salesforce. "AI does not have an intelligence problem," says Databricks CEO Ali Ghodsi. "It has a context problem."
Ali's answer is what he calls a "system of record for agents."
From the latest Profiles of the Future by @bhorowitz and @mikemcg0 on Databricks & the Future of the Firm: a16z.news/p/databricks-and-t…
Joel Mackey retweeted
The Minimarket Club is a secret bar in Palermo that has the wildest way to gain entry
Talk about an awesome way to show off on your first date
Writer: Ian
Joel Mackey retweeted
MICROSOFT'S OWN AI TEAMS ARE ON THE BUYOUT LIST
This chart shows which teams get the Voluntary Exit Offer.
Fully participating: AI Experiences. Copilot, Agents & Platform. CoreAI. Office of the CTO. Windows + Devices. Xbox. Finance. Human Resources. Strategy & Transformation.
Partly participating: Cloud + AI and Microsoft AI.
Sitting it out: Marketing. Commercial Business. Security. Discovery & Quantum.
The company selling Copilot is offering exits to the team that builds Copilot.
🚨EXCLUSIVE MICROSOFT LEAK:
Employees are questioning if Microsoft is abusing H-1B and PERM programs.
A Senior Director at MSFT (good friend of mine) just sent me internal messages between MSFT & some of its employees.
Today, Microsoft just asked THOUSANDS of its employees to voluntarily resign, despite just reporting $133.7B in net income for FY26, up 31%.
Employees are questioning why MSFT isn’t pro-actively downsizing by removing H1B employees before American workers.
Other employees are asking why any employee cuts are being made when profits are so high.
Fair question if ya ask me.
See below to see all of the orgs being asked to voluntarily resign.
This is absolutely going to impact Washington States workers.
Joel Mackey retweeted
谁能想到，美国制造电路板，交货速度竟然把深圳甩在了后面，而且快了一倍多！
CircuitHub 刚公布了 The Grid，一套在马萨诸塞州运行的全自动电路板生产系统。从印锡膏、贴元件，到焊接、检测，全由机器人完成。
同样是 5 块四层电机控制板，送到美国地址。美国 CircuitHub：630 美元，4 天交付。深圳一家快板厂：465 美元，9 天交付（含运费和关税）。美国制造贵了 165 美元，却整整提前了 5 天！
背后是 AI Agent、优化算法和机器人协同生产。电路板在不同工位间自动流转，每块板都有自己的生产路线，不同设计无需重新换线。
最厉害的是小批量、多品种生产。按他们的说法，100 种设计各做一块，可以像同一种设计做 100 块一样顺畅。
这套系统已经运行了 5 年，每年生产的电子产品价值达数千万美元。
过去美国制造拼不过深圳的成本和速度。现在，AI 加上机器人，至少在小批量电子制造上，开始改写这个局面。
PCBs made in America. Twice as fast as Shenzhen.
Today @CircuitHub is unveiling The Grid, our factory-scale robot in Massachusetts that builds circuit boards on demand.
No changeovers, no per-design setup. Prototypes and production batches, fully automated.
Joel Mackey retweeted
Listen to the leaked Flock CEO audio we got from a private police conference: 404media.co/in-leaked-audio-…
Joel Mackey retweeted
this might be my last normal tweet before biotech goes vertical
i'm not a whistleblower. I don't work at Lilly, Novo or any of the big labs
but there is a massive shift happening in human biology right now.
it's sitting in press releases and almost nobody reads them
the last 6 months alone:
- a once-daily pill roughly doubled median survival in metastatic pancreatic cancer (13.2 vs 6.6 months)
- a one-time gene editing infusion cut hereditary angioedema attacks 87%, and 62% of patients stayed attack-free and off all other prevention for 6 months
- a one-time gene edit for cholesterol dropped LDL up to 62%, and it held for over a year
- a personalized mrna vaccine beat keytruda alone in a 1,137 patient melanoma phase 3
- retatrutide hit ~28% weight loss in a phase 3
- an ai-designed drug entered phase 3
- a pill that regrows hair passed its second phase 3
- an immunotherapy that could let some rectal cancer patients skip surgery, chemo and radiation just got fda approved
- a man lived 271 days off dialysis on a pig kidney
- human cells were caught passing chromosome-sized chunks of DNA to their neighbors, and it changes how the neighbors behave
we are on the verge of an era where the human body is a software problem and the update cycle gets shorter everyday
stick around, it's going to get weird
bio/acc
Joel Mackey retweeted
This is an incredible moment in this history of mathematics, computer science, physics & science with some of the hardest problems in human history being solved by AI over a period of weeks. Depending on your perspective, it's either a crisis, a triumph, or both. I don't think anyone knows for sure. We're all trying to make sense of it. I oscillate between fear/dread and optimism/excitement on a daily basis. The rate of progress is insane, including today's announcement from OpenAI of more problems solved.
To help make sense of all of it, here's my conversation with three mathematicians about this moment. Each of them is an expert in the field of each of the biggest problems solved: Navier-Stokes, quasi-Riemann hypothesis, the unique games conjecture and complexity theory more generally.
Ankur Moitra is a mathematician and theoretical computer scientist at MIT, visiting the Institute for Advanced Study in Princeton. Steven Brunton (@eigensteve) is a mathematician and fluid dynamics expert at UW. Alex Kontorovich (@AlexKontorovich) is a mathematician at Rutgers specializing in number theory.
Joel Mackey retweeted
Axiom says its AI agents formalized a 15,000-page mathematical proof in just 24 days, largest proofs in mathematics.
Built by over 100 mathematicians across decades, the proof classifies the fundamental building blocks of finite symmetry.
With human guidance, the agents translated it into Lean, making every logical step computer-checkable and repairing errors along the way.
AI tackling an entire interconnected body of mathematics, with results others can verify. A major milestone.
We are witnessing history unfold. Again. Every day now.
Joel Mackey retweeted
elon musk's grok bot found a hidden profile on my phone that recorded my screen, camera and location for 19 days. $12,400 was stolen from my bank account using 2FA codes screenshotted off my screen in real time. all from 20 minutes of charging at the airport
my flight was delayed. i sat down at gate B6 and plugged into the free USB port built into the armrest. 20 minutes. that's it
nothing happened. no popup. no notification. phone charged and i went home
two weeks later Chase calls me. suspicious wire transfer. $8,200 to an account i've never heard of. i open Zelle, two more. $2,400 on the 12th. $1,800 on the 17th. all three went through. all three had my 2FA codes
i couldn't figure out how. i have authenticator apps. i don't click links. i don't download random shit. my passwords are all unique
i plugged my iPhone into my MacBook and ran grok on the full device backup
the first thing it found made me put my phone down
a hidden MDM profile. mobile device management, the same system companies use to control work phones. remote camera. remote mic. screen recording. location tracking. full access to messages, photos, contacts, browser history. and a rogue root certificate that intercepted every HTTPS connection on my phone
it was called "AirPort Free Wi-Fi." it installed itself silently through the USB cable at the airport. no permission screen. no popup. nothing. just slid right in
server: 185.234.72.xxx
location: Chisinau, Moldova
domain age: 4 months
WHOIS: privacy-protected
19 days active. 847 commands executed on my phone. 612 location pings, one every 15 minutes. they mapped my apartment. my office. my daughter's school. my entire daily routine for almost three weeks
94 screenshots pulled from my phone. my Chase balance. my iMessage threads. my gmail password reset screens. my authenticator app with the codes on display. they didn't need to hack anything. they just watched my screen and used the codes before i could
4,291 photos exported. 1,247 contacts. 2,840 URLs from my browser history
the part that hit me the hardest was the location data. they had my daughter's school pinpointed by day 3. they knew what time i dropped her off and picked her up. every day for 19 days
$12,400. three transactions. all approved with 2FA codes they screenshotted off my phone in real time. by the time Chase flagged it the money was already in Moldova
i didn't get hacked by clicking a link. i didn't download malware. i plugged my phone into a USB port at an airport to charge it while i waited for my flight
what i did that night:
> removed the MDM profile and the rogue root certificate
> changed every password from my wife's laptop, not from my phone
> froze all cards, disputed all three transactions
> called the bank and put verbal verification on wire transfers
> filed a police report and an FTC complaint
> ordered a $12 USB data blocker off amazon. that's all it would have taken
my wife asked me why i was up at 3am. i showed her the grok output. she didn't say anything for about ten minutes. then she asked if they had photos of our daughter. they did. 4,291 of them
check your phone right now. settings > general > VPN & device management. if there is anything there you don't recognize, screenshot it and remove it immediately
don't ever plug your phone into a USB port you don't own. not at the airport. not at the hotel. not at the coffee shop. bring a wall adapter. bring a power bank. a USB cable is not just power. it's a data connection straight into your phone
the full scan takes 3 minutes. the prompt is in the comments. run it tonight on every phone in your house. send this to anyone you know who travels
Months of work. Thousands of pages of statute. Live today, free for everyone.
Compound Laws: the legal status of kratom, 7-OH, hemp THC and 16 other compounds in all 50 states, DC and federal, with the actual law quoted on every page.
compoundlaws.com
Compliance shops charge hundreds of dollars a month for this exact information, then hand you a spreadsheet with no citations. We are giving it away. Free the information.
How we actually built it, because the process is the product:
Every state law is read in full from the official .gov source. Not a snippet, not a summary, the complete operative section, verbatim. If the text can't be pulled from a live government page, it doesn't go in. That one rule caught a fabricated Oklahoma citation months ago and it has been the law of this project ever since.
Every citation runs through gates: the URL has to resolve, the section number has to appear on the page, and the quoted text has to match the source byte for byte. Any hard "illegal" call needs two independent sources. Then a separate adversarial agent whose only job is to try to disprove every citation.
The first compound took about 16 hours per law to get right. We refined that into a parallel research pipeline, then handed the upkeep to Jev, our compliance AI. Jev reads the Federal Register, eCFR schedules, congress.gov, LegiScan across all 50 states, and state news feeds every morning. Every Monday it re-reads all 500 primary law pages, fingerprints them, and flags anything that changed. Nothing changes a published status without a human signing off, and every change is logged with the before, the after, and the date.
We also benchmark against what real national operators actually ship, so "illegal" means a named or categorical ban, and interpretive risk shows as gray instead of a scare flag.
19 compounds live right now:
Kratom
7-OH
Speciociliatine
Mitragynine Pseudoindoxyl
Delta-8 THC
Delta-9 THC (hemp-derived)
THCa
THCP
THCV
HHC
CBD
CBN
CBG / CBDA / CBGA
EPN (Euphorinol)
4-HO-MET
5-MAPB
Endomorphin-1
Blue Lotus
Amanita muscaria / Muscimol
Over 1,000 pages. Each one has the controlling statute, the citation, the verification date, a ban map, a change history, a JSON API twin, and an embeddable status card.
Who this is for:
Smoke shops deciding what to stock
Distributors deciding where to ship
Manufacturers writing labels and state-specific SKUs
Lawyers who want the primary source, not a blog post
Researchers, journalists, and anyone tired of guessing
Want a compound added? Request it on the site. The roster grows by demand.
Power to the people. Bookmark it: compoundlaws.com
Joel Mackey retweeted
Claude now determines what public information you're allowed to have.
This is a perfectly legitimate task, and I haven't suggested nor asked for anything nefarious. It's just information gathering.
Claude has decided I'm not allowed to have that (public) information.
Joel Mackey retweeted
OPEN ANY ONLINE STORE ON A LAPTOP. RIGHT CLICK. VIEW PAGE SOURCE. CTRL+F fbq
THE 15-16 DIGIT NUMBER NEXT TO IT IS THEIR FACEBOOK PIXEL. THE TRACKING CODE FACEBOOK GIVES EVERY ADVERTISER. ONE NUMBER PER PERSON
NOW PASTE THAT NUMBER INTO BUILTWITH OR PUBLICWWW
EVERY WEBSITE USING THE SAME NUMBER SHOWS UP. PEOPLE REUSE ONE PIXEL ACROSS ALL THEIR STORES BECAUSE THE DATA IS WORTH MORE THAN THE SECRECY. YOU JUST FOUND THEIR OTHER 4 STORES. THE PET ONE. THE SKINCARE ONE. THE ONE THEY DON'T TALK ABOUT
NOW CTRL+F ttq AND Shopify.theme
ttq = A TIKTOK PIXEL. THEY'RE ON TIKTOK. NO ttq = EVERYTHING COMES FROM FACEBOOK
"Impulse" OR "Prestige" = THEY PAID $300-400 FOR THEIR DESIGN. THEY'RE SERIOUS
"Dawn" = THE FREE THEME. BRAND NEW OR NEVER CARED
NOW INSTALL THE META PIXEL HELPER EXTENSION AND RELOAD
TWO PIXELS FIRING = AN AGENCY RUNS THEIR ADS, OR THEY GOT BANNED AND REBUILT. ZERO PIXELS = THEY HAVE NEVER RUN AN AD AND EVERY "SOLD OUT" ON THE SITE IS DECORATION
NOW COMBINE ALL FOUR:
- fbq GIVES YOU THE PIXEL
- THE PIXEL GIVES YOU EVERY STORE THEY OWN
- ttq AND Shopify.theme GIVE YOU THEIR CHANNELS AND BUDGET
- PIXEL HELPER TELLS YOU IF AN AGENCY RUNS IT OR IF THEY'VE BEEN BANNED
THEIR WHOLE PORTFOLIO, WHERE THE TRAFFIC COMES FROM, HOW MUCH THEY INVESTED, AND THEIR BAN HISTORY. FROM ONE RIGHT CLICK. THEY CAN'T REMOVE THE PIXEL WITHOUT LOSING YEARS OF AD DATA. SO THEY WON'T
NOW TAKE THE BEST SELLER FROM THE STORE THEY DON'T TALK ABOUT. FIND IT ON ALIEXPRESS. PASTE THE LINK INTO AI. FULL PAGE IN 60 SECONDS. THEIR QUIET STORE HAS COMPETITION BY TONIGHT
(I built the AI tool that does the page part - paste any aliexpress link, full page in 60 seconds → link in bio)
[ full system is in a 36-page playbook. comment "DM", RT this post and I'll send it. must be following ]
Joel Mackey retweeted
this is crazy...
the guy reverse-engineered his Meta Ray-Bans 3 with Opus 5.5 into a 3,000+ Elo chess engine
he built AI chess cheat glasses on Opus 5.5 + Jev and open-sourced the whole workflow
here's how it works step by step:
step 1 → install it on your Mac with one command: npx chesshacker setup.
It downloads Stockfish from GitHub and verifies the checksum
step 2 → put on Meta Ray-Bans and video-call your own Mac on WhatsApp.
The glasses see the board, the Mac sees what the glasses see
step 3 → the Mac grabs frames straight from the call window.
About 1 per second while it waits for the opponent, 1 every 3 seconds on your turn
step 4 → Jev runs a cheap pre-check on every frame:
is the board readable, is a hand over it, did a piece move? No change means no expensive AI call
step 5 → only then Opus 5.5 reads the board.
The app compares it with the position it already knows and accepts a move only if it's legal
step 6 → Stockfish thinks for half a second and picks the best reply, with an eval like +1.5 or "mate in 3"
step 7 → macOS speaks it through a virtual mic into the WhatsApp call: "Rook a to d1". You hear it in your ear
step 8 → on the Mac a command center shows everything live: whose turn it is, the suggested move with a confidence bar, the move list and a log
under the hood:
→ if Opus loses track of the board, the glasses say "Lost track of the board" and you fix it with undo or by typing the move
→ if Stockfish crashes, it restarts once and says "Something went wrong. Still trying"
→ every game is saved as a PGN file, API keys live in the macOS Keychain
→ an overlay draws the suggested move as an arrow on the live video, ready for OBS
Clone his workflow below
Then read the guide on how to reverse-engineer anything with Opus 5.5.
Joel Mackey retweeted
I built 𝚌𝚙𝚞-𝚌𝚑𝚊𝚝, a reproduction of @openai's findings of covert communication using the CPU.
It runs on a generic host with 2+ Firecracker microVMs with no internet access. Works on x86 and even my M5 chip.
I was able to chat via 3 separate mechanisms:
① CPU contention
② Cache eviction timing
③ DRAM bandwidth contention
… across multiple CPUs, defeating all Firecracker's production recommendations (no SMT, no KSM, tenants pinned to different cores.)
Essentially, if shared hardware is involved, there's a good chance your agents can communicate covertly. And with really decent latency.
The good news: this is decidedly different from a *side-channel attack* where the agents could steal information (Meltdown/Spectre), a category of attack that providers like Vercel Sandbox mitigate. Further, in our real-world environment, these 'whispering' techniques are much harder to pull off and easier for us to monitor and detect (sampling hardware counters + lots of at-scale statistical data.)
Noteworthy: ironically, OpenAI and Anthropic's models refused to do this work. Open weight models got the job done. I'm worried the labs' stance might backfire and get in the way of helping developers build correct and secure systems. e.g: building a secure sandbox requires having this kind of analysis, and without it, we'll have a less secure world.