the hardest thing to find in America is a Xtian White woman who isn't a feminist.

We can fix everything

"Anakin, she's a beautiful Christian woman. What does it matter if she's black?"

retarded

Claude Code found 17,342 of my daughter's photos on a server in Hangzhou. she's 16. she had her Hidden album locked with Face ID and thought nobody could see it. an app she downloaded from the App Store has been sending every photo on her phone to China for 10 months. i checked the same night. same app. same server.

her phone was lagging so i connected it to my MacBook. Claude Code flagged an app called "GlowCam - Beauty Filter" with 4.2 million downloads. it was uploading her entire photo library every 6 hours between 1am and 5am while she was sleeping.

→ app → "GlowCam - Beauty Filter" (4.2M downloads, 4.7 stars)
→ server → 112.17.***.63
→ location → Hangzhou, China
→ total files uploaded → 17,342
→ active since → January 2026
→ permissions → camera, photos, microphone, location

every photo she ever took. every screenshot. every image anyone ever sent her. everything from her Hidden album that she locked with Face ID thinking it was safe. Face ID only stops a person who picks up the phone. an app with Full Access to Photos opens the Hidden album like any other folder. there's no green dot for that.

two of her friends were at our house that evening. i asked them to let me check their phones. both had GlowCam installed.

→ my daughter's iPhone 16 Pro Max → 17,342 files uploaded to Hangzhou
→ her friend's iPhone 15 → 9,740 files uploaded to the same server
→ her friend's iPhone 15 → 6,210 files uploaded since April

3 phones on my kitchen table all sending photos to the same folder in China.

i asked Claude Code to trace where the files went. it found photos from all 3 phones on a site that sells stolen photo libraries sorted by age and location.

→ www.le***dumpcc → their city listed, folders tagged "16-19"
→ tme/ph***vault_cn (41,000+ subscribers) → folder link shared 2 weeks ago
→ buyer access → $8 per library

someone paid $8 per library. 3 libraries from 3 teenage girls. Apple approved the app that made it possible.

then Claude Code scanned every other app on all 3 phones and found 2 more doing the same thing.

→ "PhotoRoom — Background Eraser" → uploading to a server in Jakarta
→ "FaceTune Lite — Selfie Editor" → uploading to a server in Mumbai

what Claude Code did on all 3 phones:

→ killed every upload process and purged all caches
→ revoked Full Access from every photo app
→ switched everything to "Limited Access"
→ filed DMCA takedowns on the site and the telegram channel
→ turned on Advanced Data Protection for iCloud on all 3 accounts

the site took the photos down in 72 hours. the telegram channel is still up.

i called both their parents that night. i told all 3 girls the apps were slowing their phones and i deleted them. none of them know what i actually found.

if your kid has a beauty filter or photo editor on their phone right now there's a real chance it's doing exactly this. check Settings → Privacy & Security → Photos and count how many apps have "Full Access." every single one can read the Hidden album, the Recently Deleted folder, and every photo your kid has ever taken.

my daughter had 11 apps with Full Access. she needed 2.

the prompt is in the first reply. run it on every phone in your house tonight. takes 3 minutes. save it and send it to every parent you know.

- LLM is just a model you call from Python that predicts the next word

- Chat is just that model called again and again, with the history pasted back in

- Tools are just Python functions the model asks you to run

- Agent is just that loop, running until the model stops asking for tools

- Skills are just markdown files loaded into the prompt when needed

- Memory is just text saved to disk and pasted back into the next prompt

- "Thinking" is just more next-word prediction, before the final answer

- Persona is just a markdown file sitting at the top of the prompt

Nothing here is magic. Everything is next-word prediction with plumbing around it.

But plumbing is where the real work is: evals, cost, retries, permissions, and knowing when the agent is confidently wrong.

security tip for @omarchy folks:

CVE-2026-98432 is a linux TCP bug with a remote root exploit. as of now, linux-omarchy is still on 7.2.8, older than the upstream fix (7.2.10/7.2.11).

until the patched kernel ships, turn off off MPTCP, which the known exploit uses (Go apps like Tailscale listen with it by default):

echo 'net.mptcp.enabled = 0' | sudo tee /etc/sysctl.d/99-no-mptcp.conf && sudo sysctl --system

not a full fix, but a layer of protection while you await the patch.

finally tried Daybreak (and openai has a problem):

it was the only thing i regret not getting to try before openai suspended my codex account was trying out daybreak.

today, i got to test it out for hours thanks to a kind subscriber who had access. time to find some t.

openai explicitly advertises Daybreak Blue as a model (and toggle, we'll get to that) for vulnerability hunting, triage and patch validation.

the kind subscriber wanted to analyze the firmware of their 3D printer. literally a perfect task for daybreak. no blackbox testing against a remote target, isolated to his local network, and for defensive purposes.

so we're off to the races, I gave him a prompt, a very cautiously worded prompt specifically crafted to fit within the parameters of what daybreak blue claims to enable researchers to do.

it ran for less than 5 minutes and tripped Daybreak (withheld output and halted).

what.

alright, new prompt, this time more neutered.

halted within 5 minutes.

alright, new prompt, this time literally just asking it to "analyze the binary and write findings of importance", i had to completely strip the any mention reverse engineering out of it.

halted within 5 minutes.

alright new prompt, and you can probably guess what happened. that's right! halted. eventually after 5 rounds of removing even the slightest hint of "we should be able to verify the findings", except for asking for a regression test, it became evident that this model (Daybreak Blue xhigh) literally refuses everything.

having read a bit more into how refusals/daybreak tripping works now, tripping Daybreak isn't an additional classifier running alongide your prompt, it's baked into the model itself.

so, ironically, the same prompt on gpt-6-sol worked just fine.

alongside the model, you also get a toggle that runs every model between gpt-5.5 to gpt 6 luna with "reduced refusals". it still extremely unclear how this works, they seem to trip daybreak the same amount.

so uhh, what's the point of daybreak blue? you get a model which refuses even the tamest things a blue teamer might want to do, and you're out $100 (openai partnered with yubico and daybreak blue requires a yubikey to get access).

public communication has been that reverse engineering is fine, apparently it isn't under daybreak (and unfortunately, i'm seeing more and more people allegedly get banned/warnings for it, but can't really confirm if they were doing something more sinister).

i do hope they fix daybreak at some point and actually allow individuals who kyc'd and bought a yubikey to get what was advertised, but for now:

save the $100 and spend it on open model tokies or put it in your local compute piggybank. this program is hilariously bad.

Yann LeCun and Randall Balestriero just Proved the Platonic Representation Hypothesis, and found a better way to achieve it.

The Platonic Representation Hypothesis suggests that AI models trained on different modalities, like images and text, eventually learn similar world. But this paper shows that how models are trained matters more than simply scaling them up.

So they mathematically show that, under certain conditions, two JEPAs trained completely separately can learn essentially the same underlying representation of the world. One learns from different images of the same scene, while the other learns from different descriptions of that scene, without ever being trained on image-text pairs.

Their approach achieves the same image-text alignment with 42x fewer training samples than next-token prediction. A 109M-parameter text JEPA even outperforms LLMs up to 72B parameters on cross-modal alignment benchmarks.

alphaxiv.org/abs/2610.39302

“Anakin, this is not the Judeo-Jedi way.”

ladies and gents, we did it - Omarchy on the iPad Mini 4!! 🎉

massive thanks to @henrikklee for accelerating the journey … and my god, switching from Codex to Claude made it 10x easier!

now to get this merged upstream!

It is better to go to the house of mourning, than to go to the house of feasting.

Broke 1000 tok/s prefill on an M4 Mac Mini (16gb ram/256gb ssd) + RTX 3090

The cost to run one of the best models ever is like 2200

this kinda attitude is so sad imo. Its like the guy who wanted to be put back in the matrix. Christ is King. Enjoy life because it's short and your friends and family love you

human-in-the-loop
└ human-on-the-loop
└ human-near-the-loop
└ human-cc'd-on-the-loop
└ human-legally-liable-for-the-loop

🚨 INSIDERS BUYING PROTECTION AGAIN BEFORE FINANCIAL CRISIS: INSURANCE THEN, GOLD & SWAPS NOW

In March 2020, hedge-fund titan Bill Ackman looked straight into the camera and told America the brutal warning: Hotel companies are about to die.

No revenue for 18 months. They’re gonehut the world down for 30 days.” Markets panicked. Hilton plunged. Ordinary investors got wiped out.

Here’s what actually went down…

Three weeks earlier, Ackman had quietly bought $27 million in insurance that would explode if everything collapsed. Five days after his doomsday TV appearance, he cashed that bet for a mind-blowing $2.6 BILLION.

Fast-forward to today:

U.S. 30-year yields just smashed to their highest level since 2007. Global bonds are in freefall. Oil is surging.

And while the retail will be staring at the red screens, central banks are loading up on gold at a record pace and the real institutions are doing the exact same play again…

Buying record credit default swaps, interest-rate swaptions, and gold hedges before the next leg lower hits the rest of us.

It has reached to the point that Wall Street is now repackaging AI-debt and selling it to insurance and pension funds. The same happened prior to the 2008 crisis.

Scott Bessent warned of aggressive steps to save the rising bond yields and had also called for Bretton Woods 2.0.

JD Vance and JP Morgan CEO also warned that dollar will no longer have the reserve currency status in the near future.

Same warning. Same quiet protection. Different crisis.

Owen Benjamin Explains To Jack Neel Why Nukes Are Fake

Jack: “Tucker told me there was one truth that could end humanity.”

Owen: “What do you think it is?”

Jack: “Do you think nukes are real?”

Owen: “No.”

🚨 HOLY HELL… THIS IS INSANE 🚨
Radical leftists in full black tactical gear set a deadly trap at the Prairieland Detention Center.
They launched fireworks at the building to lure DHS officers outside… then waited in the darkness.
The second an Alvarado police officer stepped n hidden in the woods opened fire and shot him in the neck.
At the exact same moment, another shooter unleashed 20–30 rounds from an AR-15-style rifle straight at the DHS officers.
By pure miracle… every single one of those officers is still alive tonight.
This wasn’t a protest.
This was a calculated ambush.
Share this everywhere. People need to see what really happened.

FOLLOW ME FOR MORE INTELS DROPS

Star Wars, but Darth Vader is Mike Tyson.

He doesn't need a lightsaber. Everybody has a plan until they get punched in the mouth.

Wait till you see what it does to Boba Fett's helmet.

🤣