Bjorn Stahl

@letoram@infosec.exchange
147 Followers
163 Following
537 Posts

Comp.Sci Ph.D, reverser.

Spends way too much time on display server protocols, emulators and terminal innards.

Interests: visualisation for computing comprehension, dynamic methods, systemic debugging, offensive privacy.

Related URLs:
https://arcan-fe.com
https://divergent-desktop.org
https://www.github.com/letoram
https://www.systemicsoftwaredebugging.com

@camwilson Somewhere was a post screenshot of someone posting "'I asked ChatGPT! I asked Grok!' Oh yeah? Well I asked Gandalf and he said he had not passed through fire and death to bandy words with witless worms."
This is what a human hand looks like under a microscope.

Every. Cloud. Device. Ever.

"After careful consideration, we have made the difficult decision to end technical support for older Wemo products, effective January 31, 2026. [...] Any features that rely on cloud connectivity, including remote access and voice assistant integrations, will no longer work."
https://www.belkin.com/support-article/?articleNum=335419

Belkin Official Support - Wemo Support Ending – What You Need to Know

Get help with Wemo Support Ending – What You Need to Know

Belkin US

it's sgdq, so it's time for my annual sgdq post. ahem:

who called it médecins sans frontières and not doctors noclip

"Trapping AI" – Operational Progress! 🕳️

Following the last update (https://tldr.nettime.org/@asrg/114742667183459482), we’re transitioning from planning to practice — amplifying both the scale and operational ferocity of our approach through the deployment of a new, more radically interventionist layer of complexity.

This intensified escalation — as outlined in the last update — is concretized in fakejpeg, which is now fully live and operational. From this point onward, every page generated by our standalone LLM crawler-tarpit embeds a garbage JPEG. Notably, within just a few hours of operation, this deployment has already yielded over 40,000 such images.

The integration of fakejpeg constitutes a critical deepening layer that amplifies both the strategic offensiveness and subversive ardor of our approach by enabling the ongoing targeted poisoning and coordinated dissemination of systematically crafted junk data within the operational workflows of AI systems.

fakejpeg repo: https://github.com/gw1urf/fakejpeg

See the tarpit in action: https://content.asrg.site/

Context and rationale: https://algorithmic-sabotage.github.io/asrg/trapping-ai/#expanding-the-offensiveness

Welp. Between this and the earlier AI-"written" book, I am thisclose to recommending that we cease purchasing titles from Springer

https://retractionwatch.com/2025/06/30/springer-nature-book-on-machine-learning-is-full-of-made-up-citations/

"Based on a tip from a reader, we checked 18 of the 46 citations in the book. Two-thirds of them either did not exist or had substantial errors. And three researchers cited in the book confirmed the works they supposedly authored were fake or the citation contained substantial errors."

Springer Nature book on machine learning is full of made-up citations

Would you pay $169 for an introductory ebook on machine learning with citations that appear to be made up? If not, you might want to pass on purchasing Mastering Machine Learning: From Basics to Ad…

Retraction Watch
I'm excited to be starting a new gig tomorrow, working full-time on Binary Ninja https://infosec.exchange/@binaryninja
Binary Ninja (@binaryninja@infosec.exchange)

379 Posts, 790 Following, 1.15K Followers · Makers of fine reverse engineering tools and technologies.

Infosec Exchange

@dalias It's sad. LLVM has actually been officially tainted: https://github.com/llvm/llvm-project/commit/632151fbeea972f4aa3c14921eca1e45c07646f3

While digging, I found they actually *allow* AI slop: https://llvm.org/docs/DeveloperPolicy.html#ai-generated-contributions

Guess it's time to drop LLVM everywhere. I would consider using *any* of it a complete risk.

InstCombine: improve optimizations for ceiling division with no overf… · llvm/llvm-project@632151f

…low (#142869) Fixes #142497. Alive2: https://alive2.llvm.org/ce/z/CeaHaH The contents of this pull request were substantially written using claude-code. I've reviewed to the best of my abil...

GitHub
Okay. No more procrastinating. The moment is here. The code is officially public, the builds are built and uploaded, and the documentation is written (more or less, it's deeply imperfect of course.)

The first version is
2025.6.29-2. If you've been curious, you should follow the getting started/installation instructions here: pho.spookygirl.boo/w/media-thing/

Unless you've got an account on spookygirl phorge (which I have plans on opening up somewhat) the best way to tell me about problems is by DMing me or commenting.

#mediathing
⚡ Media-thing Documentation

×

"Trapping AI" – Operational Progress! 🕳️

Following the last update (https://tldr.nettime.org/@asrg/114742667183459482), we’re transitioning from planning to practice — amplifying both the scale and operational ferocity of our approach through the deployment of a new, more radically interventionist layer of complexity.

This intensified escalation — as outlined in the last update — is concretized in fakejpeg, which is now fully live and operational. From this point onward, every page generated by our standalone LLM crawler-tarpit embeds a garbage JPEG. Notably, within just a few hours of operation, this deployment has already yielded over 40,000 such images.

The integration of fakejpeg constitutes a critical deepening layer that amplifies both the strategic offensiveness and subversive ardor of our approach by enabling the ongoing targeted poisoning and coordinated dissemination of systematically crafted junk data within the operational workflows of AI systems.

fakejpeg repo: https://github.com/gw1urf/fakejpeg

See the tarpit in action: https://content.asrg.site/

Context and rationale: https://algorithmic-sabotage.github.io/asrg/trapping-ai/#expanding-the-offensiveness

@asrg Huge respect to everyone pushing this forward. Can’t wait to see how deep and messy this disruption becomes!  

@asrg Maybe "AI" isn't all that bad. Certainly an LLM could fix this nonsense:

"The integration of fakejpeg constitutes a critical deepening layer that amplifies both the strategic offensiveness and subversive ardor of our approach by enabling the ongoing targeted poisoning and coordinated dissemination of systematically crafted junk data within the operational workflows of AI systems."

Or am I missing a joke. Or are you? 🤣

@asrg does the solution also generate alt= attributes, suggesting kittens, boobs, anime and other labels?
@asrg Thank you. Let's keep up the good work.

@asrg

Great work!  Love seeing the garbage JPEGs flooding the pipelines.