Florimond Houssiau

Unbelievable things coming from the handmade LLM community

youraislopbores.me

EU agrees to chop meaty names from vegetarian and vegan food products Lawmakers will ban use of 31 meat-related names as part of efforts to help livestock farmers in food supply markets

So "lawmakers will outlaw use of 31 meat-related names as part of efforts to help livestock farmers in food supply markets"... just when health, climate & environment scientists are stressing the urgency of helping more livestock farmers OUT of food supply markets?
And what a silly, pointless move!

Our new paper demonstrates that removing images of children from training datasets of text-to-image models fails to prevent the misuse of these models to generate child sexual abuse material (AI CSAM). A thread 🧵

European Parliament votes to ban use of ‘veggie burger’ and ‘plant-based sausage’ - European Vegetarian Union Brussels, 8 October 2025 European Parliament votes to ban use of ‘veggie burger’ and ‘plant-based sausage’: coalition calls decision misguided and counterproductive This Wednesday the Members of the E...

WOW! The European Parliament just voted to BAN the words 'burger' and 'sausage' when used in sustainable plant-based alternatives. HUGE win for the meat and dairy lobby in the EU, and lots more confusion for consumers if this goes through. Insane!
www.euroveg.eu/european-par...

A photo of the Nintendo switch menu where Silksong is being downloaded

Aaaaaaaaaa

A screenshot of Google's Gemini Apps activity settings, where the Turn off button is greyed out. A notification states that "This setting is not currently available for your account type", suggesting that you cannot turn off Gemini on a free account.

This is a fucking joke -- you need a paid account to turn off Gemini?? If anything this is the best proof of how deeply unpopular AI is (pay me or you must use the slop machine that costs me billions every year)

A wasteland of rubble, dust and graves: how Gaza looks from the sky The Guardian joins a Jordanian military airdrop for a rare chance to observe a landscape devastated by Israel’s offensive Seen from the air, Gaza looks like the ruins of an ancient civilisation, brought to light after centuries of darkness. A patchwork of concrete shapes and shattered walls, neighbourhoods scattered with craters, rubble and roads that lead nowhere. The remnants of cities wiped out. But here, there has been no natural disaster and no slow passage of time. Members of Jordan’s military stand among pallets of aid about to be dropped on Gaza. Continue reading...

A wasteland of rubble, dust and graves: how Gaza looks from the sky

i can’t take the hypocrisy in so much western media coverage of israel’s bombing of iran and iran’s retaliation.

israel can do whatever it wants and they’ll find a way to legitimize it, but iran is positioned as fundamentally evil and thus deserving of attack and bloodthirsty in its response.

"we need to anonymize our data so we remove names"

"My [overworked] DPO approved [this new, untested data-processing technology] so it's fine"

it's ok it's just medical data what's the worst that could happen

I posted this on the bad place, but I am once again asking privacy scientists to have a look at what's going on with medical professionals. Things are really bad there 🫠

🤢

Google trends for the search "remove meta ai whatsapp", showing a very clear peak towards the end (today), with the projection for this week being the maximum of the curve.

Welp, guess I'm not the only one that had this useless and environmentally destructive "feature" pushed to their phone. Has anyone figured out how to remove it on Android? (besides switching to Signal)

A scoping review of privacy and utility metrics in medical synthetic data - npj Digital Medicine npj Digital Medicine - A scoping review of privacy and utility metrics in medical synthetic data

My guess here is that it's a "research culture" mismatch, where the papers they reference for the closest-distance test were published in medical-adjacent journals, but I still find that unacceptable. Hopefully work like www.nature.com/articles/s41... can help redress this culture issue.

Another section of the paper: We used the partitioning membership disclosure attack method proposed by El Emam K and colleagues where, instead of using the hamming distance between samples as a similarity measure, we used a weighted Euclidean distance where the weights are defined as the entropy of each feature, as proposed by J. Yoon et al.

and instead use a simplistic closest-distance test to evaluate privacy risks of their method -- a test that is well-known to underestimate the risk!

What's baffling here is that the authors are aware of SOTA privacy work (DP, Stadler et al.) but chose to ignore it and use a much weaker notion.

A screenshot of the paper, with the following section highlighted: However, differential privacy has been shown to not fully mitigate the risk of re-identification. Stadler and colleagues have shown that under certain circumstances, neither the original implementation of PrivBayes nor PATEGAN (Private Aggregation of Teacher Ensembles Generative Adversarial Network) reliably prevents linkage attacks, leaving some samples vulnerable to membership inference attacks.

Reading synthetic data papers in healthcare journals is often infuriating as a privacy (ex-)researcher, but I stumbled across one that is really egregious. The authors cite Stadler et al. (the initial work on membership inference attacks on SDG), to (wrongly!) dismiss a SOTA notion of privacy,

What’s a technology that you think is overhyped? I’m going to give a sideways answer to this, which is that the venture capital business model needs to be understood as requiring hype. You can go back to the Netscape IPO, and that was the proof point that made venture capital the financial lifeblood of the tech industry. Venture capital looks at valuations and growth, not necessarily at profit or revenue. So you don’t actually have to invest in technology that works, or that even makes a profit, you simply have to have a narrative that is compelling enough to float those valuations. So you see this repetitive and exhausting hype cycle as a feature in this industry. A couple of years ago, you would have been asking me about the metaverse, then last year, you would have asked me about Web3 and crypto, and for each of these inflection points there’s an Andreessen Horowitz manifesto. It’s not simply that one piece of technology is overhyped, it’s that hype is a necessary ingredient of the current business ecosystem of the tech industry. We should examine how often the financial incentive for hype is rewarded without any real social returns, without any meaningful progress in technology, without these tools and services and worlds ever actually manifesting. That’s key to understanding the growing chasm between the narrative of techno-optimists and the reality of our tech-encumbered world.

Stand by this: www.politico.com/newsletters/...

And the link to the paper in question: sites.computer.org/debull/A24ju...

On a more positive note, I guess that imitation (however automated it may be) is the sincerest form of flattery -- this is really cool work (not really by me! I only helped w/ the privacy part) by @cptanalatriste.bsky.social and others (whose bsky handle I don't know) 🙌

The figures and code listings have been copied as is. Truly unbelievable that such obvious plagiarism would be published in a conference organised by IEEE.

This paper ieeexplore.ieee.org/abstract/doc..., published at ICAIC25, blatantly plagiarises a paper by myself and colleagues at the Turing,, claiming that they developed the sqlsynthgen package. Although the text has been rephrased, the contents are nearly identical.

Huh oh. As a committed anti-conspiracy-theorist for the "phones listen to your microphone and targeted ads based on what you say" thing I can tell this is going to be exhausting

This is so much better than having a liveable planet