Okay, we have a new contender for Most AI Thing to Ever Happen

1) July 25th: someone messes around with an LLM and posts a proof of the Collatz conjecture that does, in fact, verify in the theorem prover. (The AI use is not disclosed on the github page) https://github.com/xrchz/CollatzLean

2) July 26th: several serious bugs are posted in the theorem provers, that in principle could allow a false statement to be "proven" true. They're serious, yes, but no need for panic, because you're not going to blunder into accidentally exploiting the bugs while writing a proof, probably.
https://github.com/leanprover/lean-kernel-arena/pull/81

3) July 28th: someone who was right to be very skeptical of the Collatz proof, and had the expertise to study it with a fine-toothed comb, discovered it was exploiting a bug https://github.com/leanprover/lean4/issues/14576

4) The "proof" turns out to be exploiting multiple similar but distinct bugs to pass different solver variants!

5) the human who posted the proof acknowledges the AI use and claims they did not knowingly point it towards the bugs it exploited. https://leanprover.zulipchat.com/#narrow/channel/270676-lean4/topic/Counterexample.20to.20the.20Lean.20Conjecture.20.28Soundness.20Bug.29/near/613135216

Note that the proof was posted shortly before the related bug reports were posted. It is an open question if the AI found people discussing the bugs shortly before they were formally posted and "decided" to exploit them, if the AI "knew about it" as a learned strategy from the training stage (putting every single "proof" it's ever made and ever will make into profound doubt), or if it's recently been repeatedly blundering into it by sheer stupidity and that's how people noticed the bug at about the same time.

Theorem provers aren't magic, and have bugs just like all other programs. They are tools to help us double-check our reasoning. When you skip the reasoning and ask an AI to "prove" something for you that's over your head, you're entering an adversarial pact with the monkey-pawed Devil of Customer Satisfaction.

my initial source for investigating this myself: https://lipn.info/@mevenlennonbertrand/116997927457012577

@ozu

Yes: https://infosec.place/notice/B8dh429Gf6p490VuJU

And HF also didn't notice someone is messing with their k8s in the noisiest way possible for days.

This other diagram is supposed show the time distribution of events. What are considered as "events"? What is the vertical scale of this diagram? What is 461? Why do stripes contain different colors sometimes?

This diagram is also bad, but at least it shows the high-level flows.

Notice that "Third party code sandbox" is marked as "Compromised", while the post text explicitly states that "Modal’s infrastructure was not compromised in any way". Also, how do you even compromise a *sandbox*? WTF is Lateralize?!

[RSS] ColdFusion Under Fire: Breaking Down CVE-2026-48283 and CVE-2026-48313

https://horizon3.ai/intelligence/blogs/coldfusion-critical-cves/

[RSS] The Cipher Behind QSYRUPWD: Reconstructing IBM i Password Hashes

https://blog.silentsignal.eu/2026/07/28/the-cipher-behind-qsyrupwd-reconstructing-ibm-i-password-hashes/

IBM may be overselling their local password protection: my old friends show some neat #IBMi reversing tricks to reveal how the platform hashes passwords at different security levels (QPWDLVL)

HuggingFace incident report:

https://huggingface.co/blog/agent-intrusion-technical-timeline

The report itself reeks of LLM slop with gems like the "kill chain", consisting of phases like recon, exfil, c2...and k8s 🤡 Nuances are overemphasized (like how code execution was used to execute code) while important steps are blurry (e.g. they had some kind of "allowlist" in the dataset processor, that allowed everything which didn't look like a URL?).

I feel sorry for blue teams not because they'll have to respond to more incidents but because they'll have to wade through reports like this...

I'm looking for a new job.

I've been in security for 13 years. I spent the last year at Trail of Bits doing code reviews and security assessments on desktop/web apps, kernelmode code, and firmware. I co-wrote the C/C++ testing handbook we published recently, contributing the Windows usermode and kernel sections. I've got strong hardware skills and industrial experience.

UK based but flexible on operating hours. Looking for similar work, full-time remote.

CV: https://poly.nomial.co.uk/graham_sutherland.pdf

#fedihire

So Claude apparently creates a “secret” URL for the conversations people have with it.

URLs can be indexed if they can be reached. So users’ private conversations with Claude, predictably, started showing up in Google search results.

“But but but we had a robots.txt file!” says Anthropic, unironically.

https://www.wired.com/story/private-claude-chats-exposed-in-google-and-bing-search-results/

#noai #ai #consent #badideas

Hacker Wipes Romania's Entire Land Registry Database https://it.slashdot.org/story/26/07/20/172249/hacker-wipes-romanias-entire-land-registry-database?utm_source=rss1.0mainlinkanon

"This meeting could have been an email"

This meeting would have been an email if you had responded to your email

@ljrk @drwhax Re: 1., the usual question we got from CISO and above on our reports is "how does it look compared to similar companies?". I agree that this is in part the "you don't have to outrun the bear" logic, but also that people in position don't want to look incompetent in front of their peers.

Another thing is that you don't get fired if you didn't follow the hackers advice, but you do get fired if you don't pass compliance which is one part BS, and the other part is easy to cheat.

Some people will tell you that PQC resilience is a maths issue, but I'm arguing it's a hygiene issue.

@drwhax finding vulnerabilities can be stochastic because it has a very clear and effective verification function: either the exploit works or it does not. Exploit code can be messy and convoluted, as it is not going to be maintained after the vulnerability is fixed.

Vibe-coding fixes does not have that kind of verification function: the fix must not only close the specific vulnerability, but *also* not introduce new ones or re-introduce old ones, and it has to be maintainable in the future.

Dozens of Robotaxi Riders Are Falling Asleep, Sparking Frantic Calls For Emergency Services https://tech.slashdot.org/story/26/07/19/236218/dozens-of-robotaxi-riders-are-falling-asleep-sparking-frantic-calls-for-emergency-services?utm_source=rss1.0mainlinkanon

The PSF is hiring a Security Developer! Help triage vulnerabilities in CPython, fight malware/supply-chain attacks on PyPI, and build tools to keep the #Python ecosystem safe for millions of users 🐍🔒 This is a global, remote, 1 year term role, with the possibility of renewal. #Security

Apply today:
https://pythonsoftwarefoundation.applytojob.com/apply/ei03ut60y4/Security-Developer

@lorenzofb No dox plz!

NEW: I delved into the mystery of Phineas Fisher, probably the most prolific and public hacker never to have gotten caught.

This is what we know about the infamous hacktivist and their spectacular hacks against spyware makers FinFisher and Hacking Team.

There will be even more in my upcoming book.

https://techcrunch.com/2026/07/25/the-hacker-who-humiliated-spyware-makers-and-was-never-caught/

@drwhax Maybe the advice of the wiser among us will be heard after all these decades: focus on robust mitigations and attack surface reduction, because the moles can't be whacked anymore.