**"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:

GitHub

GitHub - gw1urf/fakejpeg: Generate files that are almost JPEGs with random data. Possibly useful in feeding aggressive web crawlers.

Generate files that are almost JPEGs with random data. Possibly useful in feeding aggressive web crawlers. - 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

**"Trapping AI" – Slight Update!** 🌀 Activity in the **"Trapping AI"** project is accelerating: in just under a month, over **26 million requests** have hit our tarpit URLs 🕳️. Vast volumes of meaningless content were devoured by AI crawlers — ruthless digital leeches that relentlessly scour and pillage the web, leaving no data untouched. In the coming days, we’ll roll out a new layer of complexity — amplifying both the *intensity* and *offensiveness* of our approach. This escalation builds on fakejpeg, a tool developed by @Alun Jones. 🖼️ fakejpeg generates fake JPEGs on the fly. You "train" it with a collection of existing JPEGs, and once trained, it can produce an arbitrary number of things that *look* like real JPEGs — perfect for feeding aggressive web crawlers junk 🗑️. Explore fakejpeg:

GitHub

GitHub - gw1urf/fakejpeg: Generate files that are almost JPEGs with random data. Possibly useful in feeding aggressive web crawlers.

Generate files that are almost JPEGs with random data. Possibly useful in feeding aggressive web crawlers. - gw1urf/fakejpeg

Learn more about *"Trapping AI"*: https://algorithmic-sabotage.github.io/asrg/trapping-ai/#expanding-the-offensiveness See the tarpit in action: https://content.asrg.site/

**AI is Dehumanization Technology** A call to reject the deployment and use of AI systems. Interesting piece by @npub1tz72...za03 > AI systems reproduce bias, cheapen and homogenize our social interactions, deskill us, make our jobs more precarious, eliminate opportunities to practice care, and enable authoritarian modes of surveillance and control. Deployed in the public sector, they undercut workers' ability to meaningfully grapple with problems and make ethical decisions that move our society forward. These technologies dehumanize all of us. Collectively, we can choose to reject them. 📰 Read it here:

AI is Dehumanization Technology

A call to reject the deployment and use of AI systems in Canada's public sector.

**"Trapping AI" – New Update!** 🌀 Web crawlers play a central role in the escalating race to develop ever-more powerful AI models: they tirelessly scour the web, harvesting vast quantities of content to feed large language models. *Babble* (

Personal Forgejo instance

babble

Junk food for your local LLM

), originally developed by @Joshua Barretto, is a tool that lures these crawlers into an endless labyrinth—feeding their insatiable hunger for data with masses of pointless content. To deliberately drain the resources of exploitative crawlers—and push AI models further toward collapse—we’ve deployed *Babble*, initially enhancing it through subtle, targeted modifications that align its functionality tightly with our strategic and operational priorities. *Babble* dynamically generates an unending stream of deterministic bollocks, trapping crawlers on a single site where they endlessly navigate an ever-growing sea of pages 🌊—each filled with vast amounts of useless text and dozens of links that draw them ever deeper into the tarpit 🔁. In the coming period, we’ll continue to develop and escalate this tactic—adding new layers of complexity and increasing both the intensity and offensiveness of the approach. At the next stage, we’ll openly share our code—allowing others to deploy, adapt, and build upon it. See it in action: https://content.asrg.site/ P.S. If you're looking for similar tools and frameworks to deploy or explore further, check out our list titled *"Sabot in the Age of AI"*—a record of strategically offensive methods and purposefully orchestrated tactics for facilitating (algorithmic) sabotage, framework disruption, and intentional poisoning. Explore it here: https://algorithmic-sabotage.github.io/asrg/posts/sabot-in-the-age-of-ai/

Sabot in the Age of AI Here is a curated list of strategies, offensive methods, and tactics for (algorithmic) sabotage, disruption, and deliberate poisoning. 🔻 iocaine The deadliest AI poison—iocaine generates garbage rather than slowing crawlers. 🔗

Cookie monster!

🔻 Nepenthes A tarpit designed to catch web crawlers, especially those scraping for LLMs. It devours anything that gets too close. [@aaron](

zadzmo.org

Aaron, @aaron@zadzmo.org

I work in tech: unix (any) sysadmin. One of those weirdos that actively likes managing DNS - if you say "It's always DNS," I'm looking for ...

) 🔗

Nepenthes - ZADZMO.org

Making web crawlers eat shit since 2023

🔻 Quixotic Feeds fake content to bots and robots.txt-ignoring #LLM scrapers. @marcusb 🔗

Quixotic

🔻 Poison the WeLLMs A reverse-proxy that serves diassociated-press style reimaginings of your upstream pages, poisoning any LLMs that scrape your content. @Mike Coats 🏴󠁧󠁢󠁳󠁣󠁴󠁿🇪🇺🌍♻️ 🔗

Codeberg.org

poison-the-wellms

A reverse-proxy that serves diassociated-press style reimaginings of your upstream pages, poisoning any LLMs that scrape your content.

🔻 Django-llm-poison A django app that poisons content when served to #AI bots. @Fingel 🔗

GitHub

GitHub - Fingel/django-llm-poison: A django app that poisons content when served to AI bots. ☠️

A django app that poisons content when served to AI bots. ☠️ - Fingel/django-llm-poison

🔻 KonterfAI A model poisoner that generates nonsense content to degenerate LLMs. 🔗

Codeberg.org

konterfai

konterfAI is a model-poisoner for LLM (Large Language Models) to generate nonsense("bullshit") content suitable to degenerate these models.

Let's talk about not talking about #Gaza To the Chaos Computer Club, and to the larger community that makes the Chaos Communication Congress happen. Congress has been a place for policitized hackers from all around the world to meet, share and learn. Yet this year's #38C3 event has been lived by many of us as an awkward space of silencing and self-censoring with regards to the Palestinian #genocide. We couldn't help but feel terrified at that, although not completely surprised. It has been dreadful to once again hear discussions in which the only "argument" issued was "it's a complex situation" when trying to face the harsh reality of a 2 million people country being destroyed, blocked and bombed with barely any food or clean water, while one of the last running hospital was bombed three days before. As an heterogeneous group of people that are part of communities that are interested, active or impacted by the struggle for liberation of the Palestinian people, we've felt necessary to start this discussion. What follows is a series of experiences, questions and reflections that have been shared in the self-organized session "Let's talk about the not talking about Gaza" that was announced on the last day of the Congress. ⟶

paste.aquilenet.fr

paste.aquilenet.fr

Visit this link to see the note. Giving the URL to anyone allows them to access the note, too.