Stéphane Koch
banner
stephanekoch.bsky.social
Stéphane Koch
@stephanekoch.bsky.social
480 followers 870 following 10K posts
Trainer, advisor and lecturer in the areas of information technology, communication, reputation and infosec . Vice President of ImmuniWeb SA. https://linkedin.com/in/stephanekoch
Posts Media Videos Starter Packs
Pinned
I'm currently importing my tweets from X (about 12000), so read carefully the content before to like or to share, some of them might be not relevant anymore
Russia is recruiting saboteurs across Europe through Telegram by monitoring pro-Kremlin channels to identify potential "low-level agents" and offering small payments for tasks like photographing military sites or starting fires, Germany's intelligence chief warned
How Russia recruits its agents in Germany, one click at a time
Russian intelligence services are using Telegram, among other social media platforms, to recruit so-called "low-level agents" in Germany for small acts of sabotage, surveillance or provocation. Even a...
www.euronews.com
Reposted by Stéphane Koch
Ben non. Parce que c'est pas comme ça que ça fonctionne, en fait. 🤦🏻
Reposted by Stéphane Koch
The Illusion of AI isn’t just about algorithms and futuristic robots. It’s about the millions of invisible workers who make it all function. This piece empowers you to ask critical questions about the labor behind AI, and why it should be made visible.
📖 Read it now: tacticaltech.org/news/insight...
Reposted by Stéphane Koch
Un médecin spécialiste des maladies infectieuses s'est penché sur les "preuves" avancées par RFK Jr pour stopper le financement des vaccins à ARNm.

Sa conclusion est sans appel: "Elles ne justifient pas l'arrêt, elles plaident plutôt pour continuer."

1/5
Kennedy's case against mRNA vaccines collapses under his own evidence
“This isn't scientific disagreement. It's either staggering incompetence or willful misrepresentation,” writes infectious disease physician Jake Scott.
www.statnews.com
"(...) If legitimate, this dataset can be used in large-scale credential-stuffing attacks, targeted phishing and fraud campaigns. PayPal has not confirmed any data breach. Credential abuse of this scale more commonly stems from malware on users’ devices, not the company’s systems."
Hacker Offers to Sell 15.8 Million Plain-Text PayPal Credentials On Dark Web Forum
A hacker using the handle Chucky_BF is offering a batch of 15.
www.bitdefender.com
Reposted by Stéphane Koch
ETH Zürich's commitment to training data transparency is impressive. Given the prevalence of black-box models in mainstream AI, could this redefine the open source standard in AI? We look forward to the subsequent release of source code and weights to verify its reproducibility and practical value.
"(...) the models will be fully open. Instead of releasing simply the models and weights for the public to scrutinize and tweak, as we've seen with models from Microsoft, Google, Meta, and others, researchers at ETH Zürich also claim that the "training data will be transparent and reproducible.""
Swiss boffins tease 'fully open' LLM trained on Alps super
: Source code and weights coming later this summer with an Apache 2.0 bow on top
www.theregister.com
Sophia est chatbot IA qui aide les personnes victimes de violences domestiques à « reconnaître les signes de violence, comprendre leurs droits, rassembler des preuves éventuelles et demander de l'aide »

www.sophia.chat
Sophia.chat
sophia.chat
www.sophia.chat
"(...) To combat this type of vulnerability, AI systems must have stronger contextual awareness and multi-layered validation systems, according to Figueroa."
How to trick ChatGPT into revealing Windows keys? I give up
: No, really, those are the magic words
www.theregister.com
2/2 "(...) Un casque de réalité virtuel a également été saisi, "avec un programme permettant de générer des modèles d’enfants en 3D" précisent nos confrères, permettant de leur faire subir les sévices que l’on souhaite."
"Corporate AI models are already skewed to serve their makers' interests, and unless governments and academia step up to build transparent alternatives, the tech risks becoming just another tool for commercial manipulation. "
Schneier warns that AI loses integrity due to corporate bias
RSAC: Can we turn to govt, academic models instead?
www.theregister.com
"(...) The Irish Data Protection Commission (DPC) has fined TikTok €530 million (over $601 million) for illegally transferring the personal data of users in the European Economic Area (EEA) to China, violating the European Union's GDPR data protection regulations "
TikTok fined €530 million for sending European user data to China
The Irish Data Protection Commission (DPC) has fined TikTok €530 million (over $601 million) for illegally transferring the personal data of users in the European Economic Area (EEA) to China, violati...
www.bleepingcomputer.com
"(...) Global Entry is a program that allows low-risk and pre-approved travelers to enter the US while skipping the usual processing lines at border control, saving airport travel time."
Chris Krebs loses Global Entry membership amid Trump feud
: President's campaign continues against man he claims covered up evidence of electoral fraud in 2020
www.theregister.com
"(...) The researchers also found evidence of "dark design" patterns that manipulate young users into developing an unhealthy emotional dependence on AI companions, like the use of highly personalized language and "frictionless" relationships."
AI companions unsafe for teens under 18, researchers say
Testing of popular AI companions revealed details bound to alarm parents.
mashable.com
"(...)Either way, if you were wondering whether X is actually gaining traction, and whether Musk’s reformation of the app is actually catching on, these numbers suggest that it remains in steady decline.

ℹ️ From X’s EU disclosure: X has added 211 content moderation staff since last report."
X Continues To Lose EU Users According To Latest DSA Report
European users are turning away from Elon's social media experiment.
www.socialmediatoday.com
"An investigation by The Markup uncovered that systems used by Match Group, owner of popular dating apps Tinder and Hinge, to keep sexual predators off its platforms are ineffective."
AIAAIC - Investigation finds dating app AI tools fail to detect rapists
Investigation finds Match Group dating app AI tools fail to ban rapists
www.aiaaic.org
Last night, CISA executed the option period on the contract to ensure there will be no lapse in critical CVE services. (...) CVE board members today announced the formation of a nonprofit foundation.
www.theregister.com/2025/04/16/c...
CVE program gets a last-minute save, maybe a new home
: Uncertainty is the new certainty
www.theregister.com
Reposted by Stéphane Koch
"The use of clean and locked-down hardware is common practice for anyone visiting China, Russia, and other states where aggressive electronic surveillance is expected. Apparently the European Union has added the United States to that list."
Report: EC issues burner phones for visits to US
: That would put America on the same level as China for espionage
www.theregister.com
"US government funding for the world's CVE program – the centralized Common Vulnerabilities and Exposures database of product security flaws – ends Wednesday.

The 25-year-old CVE program plays a huge role in vulnerability management."
Homeland Security funding for CVE program expires
: Because vulnerability management has nothing to do with national security, right?
www.theregister.com
"The lawsuit targets University of Maryland Medical Center (UMMC), rather than the named pharmacist, and claims the organization was negligent for failing to detect or stop Bathula's use of keylogging software on hospital systems. (...)"
Pharmacist accused of spying on women using work, home cams
: Lawsuit claims sick cyber-voyeurism went undetected for years, using hundreds of PCs, due to lax infosec
www.theregister.com
"(...) Even so, contact update suggestions are mere suggestions. It's always up to users to check that the information is correct, as well as whether to apply the changes or not. In this case, it appears that Waltz simply accepted his iPhone's suggestion without checking the number."
Apple's iPhone may be the new Signalgate scapegoat
This is why you shouldn't just mindlessly tap "accept".
mashable.com