Politechs

Good feedback, thank you.

I was very much using "AI Hater" in the @anthonymoser.com-ian sense of the word (anthonymoser.github.io/writing/ai/h...), but I totally see how this could be read in the way you describe.

We are exactly looking for folks who are informed and (rightly) sceptical.

Blood in the Machine "The most important book to read about the AI boom" (Wired): The "gripping" (New Yorker) true story of the first time machines came for human jobs—and how...

For more on the original Luddite movement, @bcmerchant.bsky.social has written the authoritative account, "Blood in the Machine". www.hachettebookgroup.com/titles/brian...

I'm using Luddite in the historically accurate sense of the word: a highly skilled tech worker who opposed deploying "all machinery hurtful to commonality".

Totally agree.

Awesome! I DM'd you, and then fixed our DMs so they're actually open now. 🤦🏼‍♂️

If anyone has a grade 1-5 teacher in your network who is a Luddite or AI Hater and who would be willing to be interviewed for Season 2 of the podcast, please put them in touch! Our DMs are open. The guest can be anonymous, as we know that it can be dangerous to speak out against AI in these times. 😞

Politechs - Teacher Email Template Template for emailing your child's teacher to discourage use of ChatGPT

If you have children in school, it's almost certain that a teacher is using ChatGPT in some way. Consider using this template, written by @anthonymoser.com, to send an email to your children's teacher(s) to voice your concerns about using AI in the classroom:

politechs.dev/teacher-temp...

Agentic ProbLLMs: Exploiting AI Computer-Use and Coding Agents This talk demonstrates end-to-end prompt injection exploits that compromise agentic systems. Specifically, we will discuss exploits that ...

One of the big takeaways from this talk is that hidden text can drive AI agent behaviour and I'm very much a fan of using this to (further) poison them.

media.ccc.de/v/39c3-agent...

Home - Academic Audio Transcription Ltd Get high-quality, human-made transcripts and closed captions from our specialist disabled-led and disabled-majority team of expert transcribers.

If you want high-quality, ethically-made transcripts, I recommend the wonderful people over at @aat-transcribes.bsky.social! They employ professional staff with years of experience, and will work with customers to find a solution that fits their budget.

academicaudiotranscription.com

AI layoffs are looking more and more like corporate fiction that's masking a darker reality, Oxford Economics suggests | Fortune "Firms don't appear to be replacing workers with AI on a significant scale," the firm said. It suspects some are trying to "dress up layoffs" as good news.

I had assumed using AI to justify layoffs was a straightforward weapon against labour, but this article suggests it may also be a signal to the market that layoffs aren't due to incompetence, but rather due to exciting technology deployed by forward-thinking executives.

fortune.com/2026/01/07/a...

This piece asks what we'll do with all the excess computing infrastructure once the AI bubble finally bursts. Quoth IBM's CEO: "no way you’re going to get a return on that because $8 trillion of CapEx means you need $800 billion profit just to pay for the interest"

www.evalapply.org/posts/after-...

There's more information in Spanish here:

tunubesecamirio.com/2026/01/21/n...

Politechs - News Politechs news feed

You can read the full press release in English here:

politechs.dev/news.html#ne...

Frena a Amazon, defiende Aragón: Contencioso histórico contra los megacentros de datos No es una "nube", sino una amenaza. Los gigantescos centros de datos de Amazon (AWS) están secuestrando el agua, la energía y el futuro de Aragón

Aragón defies Amazon: Historic dispute against mega datacentres

It's not a "cloud," it's a threat. Amazon's gigantic data centers are hijacking Aragón's water, energy, and future.

TuNubeSecaMiRio have launched a crowdfunding campaign to support legal action:

www.migranodearena.org/reto/frena-a...

Friction and frustration build resilience, and without resilience, any of the myriad setbacks that life inevitably serves up becomes a major tragedy.

The overall idea of being in the moment and giving yourself space to be with your thoughts isn't new, but I love the challenge to the tech narrative of removing friction as an unalloyed good.

In 2026, We Are Friction-Maxxing In the face of technologies that sell us escapism at every turn, we need to build up tolerance for “inconvenience.”

Just like many of the tech narratives, removal of friction is seen as an end unto itself, and becomes this unchallenged dogma (like efficiency, growth, etc), so I really enjoyed this piece by @kjezermorton.bsky.social advocating *for* friction!

www.thecut.com/article/broo...

Oops, didn't realise @adnanhusain.bsky.social was also on Bluesky. 😅

Office Hours 16-Haunting Futures: Big Tech and Resistance w/Syed Mustafa Ali YouTube video by Adnan Husain

I loved this conversation with @drsyedmustafaali.bsky.social on the Adnan Husain Show! 🔥

www.youtube.com/live/GEZQXb8...

Cars gobbling up your data and showing ads | AdGuard Ads have been creeping onto infotainment screens across various car brands. But this still cutting-edge tech increasingly reveals its darker side: continuous data collection, eroding privacy, and, ine...

Modern cars are rolling surveillance platforms, collecting data on the occupants of the vehicle and everything else within the range of their sensors. And wherever surveillance goes, advertising isn't far behind. 🤬

adguard.com/en/blog/cars...

Or to put it another way: bsky.app/profile/rhia...

Elon Musk's Grok AI floods X with sexualized photos of women and minors When contacted by Reuters for comment by email, xAI replied with the message "Legacy Media Lies."

Oh, how much should I pretend to be surprised by this completely foreseeable consequence of how this technology is being deployed? 🙄

www.reuters.com/legal/litiga...

Former EU commissioner and activists barred from US in attack on European tech regulators State department accuses group of pressuring tech firms to censor or suppress American viewpoints through regulation of disinformation

The overt actions of the Trump regime are making its authoritarian nature hard to deny. This is politechs in its rawest form.

www.theguardian.com/technology/2...

This piece from Cory Doctorow (which is the script of a speech he recently gave) on understanding the AI bubble collects a bunch of ideas he's been developing in his blog and wraps them up in one great package. It's a 20 minute read, but worth every minute of that!

pluralistic.net/2025/12/05/p...

These datasets look fine, for example: bsky.app/profile/poli...

There are datasets consisting of properly licensed material. Those are fine. Scraping random YouTube channels isn't fine with me, because the people making videos there may not be OK with their work being used to train an LLM.

This looks like a great option! I've reached out to them to explore pricing. Thanks for the recommendation! 💜

So yeah, the Gemma training datasets are not public, only slicing configs.

So OLMoASR is right out, as data scraped from the open internet doesn't meet my bar for explicit consent, unless they describe the dataset in detail and show they're only scraping correctly licensed material (i.e. with Creative Commons or similar licences).

allenai/OLMoASR-Pool · Datasets at Hugging Face We’re on a journey to advance and democratize artificial intelligence through open source and open science.

It gets worse:

"OLMoASR-Pool is a web-scale audio-text dataset collected from the public internet, consisting of approximately 3M hours of audio and 17M transcripts."

huggingface.co/datasets/all...