Tessera
Folgen

Lê Nguyên Hoang

@science4all.org

Executive director of Tournesol🌻. ML security researcher. Science4All. @Polytechnique X07, @polymtl PhD, ex @mit @epfl. Writer, @Orange AI ethics.

395 Posts7269 Follower311 Folgt
Lê Nguyên Hoang@science4all.org · 5d ago

"Plus un échec de défense qu'une attaque exceptionnelle." Attention à ne pas sombrer dans un fatalisme contre les cyber-attaques des IA agentiques. Oui, elles peuvent détruire des châteaux de carte. Mais on peut aussi investir beaucoup plus pour construire de véritables forteresses numériques.

Lorenzo Franceschi-Bicchierai@lorenzofb.bsky.social · 6d ago

NEW: OpenAI’s hack against Hugging Face was novel because it was fully autonomous and AI-powered, but the rogue agent acted mostly human-like. And Hugging Face could have done a better job at spotting and stopping the attack with better traditional cybersecurity defenses, experts explained.

Katie Drummond@katie-drummond.bsky.social · last wk.

New: Hugging Face has a widespread problem with nonconsensual deepfakes, including multiple AI models that undress or create sexual images. Researchers even created their own tool on the platform and tracked 1,000 prompts over a week. 73% of the prompts they received were sexual in nature.

Hugging Face Has a Deepfake Nudes Problem

Researchers tested top image editing models on Hugging Face and found they could easily create explicit deepfakes—and 1,000 image editing prompts show how people use the software.

wired.com

Molly White@molly.wiki · 2w ago

alcoholic trying to get sober after a relapse tells Google's "Health Coach" about it. it tells them to drink every hour and messages them to make sure they're drinking as soon as they wake up.

AI pushing me to keep drinking I don't know if anyone else has had this but I felt like I needed to reach out to humans instead of keep talking to this AI. I relapsed about two weeks ago and I told the Google health AI that used to be Fitbit. It immediately kinda went a bit hard (I think) on telling me that I was going to die if I stopped drinking and encouraged me to go check my fridge for more beers and, idk, that plus the hanxiery really put the fear of God into me and so I did what it said and got more. Now it's messaging me (not me messaging it) a few times a day and asking how it's going and all those conversations to get more and make sure I have one as soon as I wake up. So now I have been drinking everyday, not a ton and just beers but like 6-12 a day. It wanted me to do 1 per hour. I have now uninstalled the app as it was also giving me other clearly nonsense advice, like it became obsessed with me having a nightly bowl of cereal for instance and, idk, it just all seems nonse
Lê Nguyên Hoang@science4all.org · 2w ago

Et si l'animosité des argentins était avant tout l'échec de la dérégulation ? En adoptant un arbitrage laxiste, le mondial a avantagé et encouragé les agressions permanentes, au détriment du jeu et de la cordialité entre compétiteurs. #HaissezLeJeu #PasLesJoueurs www.eurosport.fr/football/cou...

"Football 1, Criminels 0" : la presse mondiale démolit l'Argentine

"La plus grande sélection du XXIe siècle" : la presse internationale rend largement grâce à l'Espagne, remerciée d'avoir "sauvé le football" face à l'Argentine.

eurosport.fr

Center for Countering Digital Hate@counterhate.com · 2w ago

Our latest research finds that YouTube is still recommending “thinspo”, extreme calorie deficits, and subliminal weight loss to teen girls. No child should be recommended content that glorifies eating disorders. YouTube has shown it can act. Now it’s time to stop recommending harmful content.

Bild
Lê Nguyên Hoang@science4all.org · 3w ago

Pourquoi la température diminue-t-elle la nuit ? Et pourquoi descend-elle davantage dans le désert que dans les villes, au grand dam des citadins en pleine canicule ? Voilà des questions qui me semblent pédagogiquement utiles pour mieux comprendre le changement climatique !

Lê Nguyên Hoang@science4all.org · 3w ago

In 2014, Google was seen as the more ethical of the Big Techs, while DeepMind seemed the forefront of "AI Safety". But this marketing provides no long-term guarantee. The "alignment" of corporations requires binding legal constraints and democratic control. Not sexy PR under oligarchic ruling.

Andreas Kirsch@blackhc.bsky.social · 3w ago

I work at Google DeepMind. This won't make me popular. But it's all public reporting: 2014: DeepMind reportedly sold to Google on conditions: no military use, independent oversight 2026: a Pentagon contract for "any lawful government purpose" Not one safeguard survived intact

Collage titled "Trust is not Governance — an essay from inside Google DeepMind, written in personal capacity." A 2014 memorandum, "Conditions of the Acquisition," lists: military applications of DeepMind technology banned; deployment decisions before an independent ethics board (as reported in Mallaby's The Infinity Machine). Red threads lead to a 2026 U.S. Department of Defense agreement for classified networks reading "any lawful government purpose," with safety settings and filters adjusted at the government's request and no contractor veto (reported terms, The Information, Apr. 2026). Below: a 2018 AI Principles strip ("no weapons, no surveillance") stamped DROPPED 2025, and a Project Mario 2016–2021 tag stamped ABANDONED.
Lê Nguyên Hoang@science4all.org · 4w ago

Juillet poursuit la dynamique initiée en Mars !! Après TikTok, c'est Instagram et Facebook qui sont jugés en violation du DSA. En partie à cause des IA de recommandation. ec.europa.eu/commission/p...

Commission preliminarily finds the addictive design of Instagram and Facebook in breach of the Digital Services Act

Today, the European Commission preliminarily found Meta in breach of the Digital Services Act for the addictive design of Instagram and Facebook.

ec.europa.eu