#c4dt_epfl — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #c4dt_epfl, aggregated by home.social.
-
#weeklyPick #C4DT_EPFL
Les récents événements autour des LLM montrent que les entreprises ne veulent pas, ou ne peuvent pas, contenir le génie qu'elles ont libéré. Face à cela, je compte sur un véritable contre-poids de la part de nos gouvernements. On observe une tendance à tout autoriser au nom du profit, alors que l'expérience a montré à maintes reprises que cela ne mène pas à un usage responsable de la technologie. -
#weeklyPick #C4DT_EPFL
Les récents événements autour des LLM montrent que les entreprises ne veulent pas, ou ne peuvent pas, contenir le génie qu'elles ont libéré. Face à cela, je compte sur un véritable contre-poids de la part de nos gouvernements. On observe une tendance à tout autoriser au nom du profit, alors que l'expérience a montré à maintes reprises que cela ne mène pas à un usage responsable de la technologie. -
#weeklyPick #C4DT_EPFL
Les récents événements autour des LLM montrent que les entreprises ne veulent pas, ou ne peuvent pas, contenir le génie qu'elles ont libéré. Face à cela, je compte sur un véritable contre-poids de la part de nos gouvernements. On observe une tendance à tout autoriser au nom du profit, alors que l'expérience a montré à maintes reprises que cela ne mène pas à un usage responsable de la technologie. -
#weeklyPick #C4DT_EPFL
Les récents événements autour des LLM montrent que les entreprises ne veulent pas, ou ne peuvent pas, contenir le génie qu'elles ont libéré. Face à cela, je compte sur un véritable contre-poids de la part de nos gouvernements. On observe une tendance à tout autoriser au nom du profit, alors que l'expérience a montré à maintes reprises que cela ne mène pas à un usage responsable de la technologie. -
#C4DT_EPFL #WeeklyPick Ca me réjouit que la confédération Suisse décide de continuer le programme BOSS pour passer de la théorie à l'action. BOSS était un projet pour voir si c'était possible d'utiliser une alternative à Microsoft Office. Maintenant la confédération veut tester ce système en vrai et avoir 3000 places de travail configuré avec OpenDesk fin 2027. Je suis très content que la confédération fait des actes pratiques, et je me réjouis déjà de voir le résultat.
-
#C4DT_EPFL #WeeklyPick Ca me réjouit que la confédération Suisse décide de continuer le programme BOSS pour passer de la théorie à l'action. BOSS était un projet pour voir si c'était possible d'utiliser une alternative à Microsoft Office. Maintenant la confédération veut tester ce système en vrai et avoir 3000 places de travail configuré avec OpenDesk fin 2027. Je suis très content que la confédération fait des actes pratiques, et je me réjouis déjà de voir le résultat.
-
#C4DT_EPFL #WeeklyPick Ca me réjouit que la confédération Suisse décide de continuer le programme BOSS pour passer de la théorie à l'action. BOSS était un projet pour voir si c'était possible d'utiliser une alternative à Microsoft Office. Maintenant la confédération veut tester ce système en vrai et avoir 3000 places de travail configuré avec OpenDesk fin 2027. Je suis très content que la confédération fait des actes pratiques, et je me réjouis déjà de voir le résultat.
-
#C4DT_EPFL #WeeklyPick Ca me réjouit que la confédération Suisse décide de continuer le programme BOSS pour passer de la théorie à l'action. BOSS était un projet pour voir si c'était possible d'utiliser une alternative à Microsoft Office. Maintenant la confédération veut tester ce système en vrai et avoir 3000 places de travail configuré avec OpenDesk fin 2027. Je suis très content que la confédération fait des actes pratiques, et je me réjouis déjà de voir le résultat.
-
This is a very thorough description of a bug detected in a Zero Knowledge Proof (ZKP) library used by a blockchain. If you are interested in how modern ZKPs work, this article gives a condensed overview of what is happening in a ZKP, and what the security guarantees are. As always, the devil is in the details, and in this case, a missing verification allowed an attacker to prove anything they wanted.
https://osec.io/blog/2026-04-30-unverified-evaluations-dusk-plonk/
-
This is a very thorough description of a bug detected in a Zero Knowledge Proof (ZKP) library used by a blockchain. If you are interested in how modern ZKPs work, this article gives a condensed overview of what is happening in a ZKP, and what the security guarantees are. As always, the devil is in the details, and in this case, a missing verification allowed an attacker to prove anything they wanted.
https://osec.io/blog/2026-04-30-unverified-evaluations-dusk-plonk/
-
This is a very thorough description of a bug detected in a Zero Knowledge Proof (ZKP) library used by a blockchain. If you are interested in how modern ZKPs work, this article gives a condensed overview of what is happening in a ZKP, and what the security guarantees are. As always, the devil is in the details, and in this case, a missing verification allowed an attacker to prove anything they wanted.
https://osec.io/blog/2026-04-30-unverified-evaluations-dusk-plonk/
-
This is a very thorough description of a bug detected in a Zero Knowledge Proof (ZKP) library used by a blockchain. If you are interested in how modern ZKPs work, this article gives a condensed overview of what is happening in a ZKP, and what the security guarantees are. As always, the devil is in the details, and in this case, a missing verification allowed an attacker to prove anything they wanted.
https://osec.io/blog/2026-04-30-unverified-evaluations-dusk-plonk/
-
#C4DT_EPFL #WeeklyPicks Google startete schon im 2010 ein 'Vulnerability Reward Program', ein Bug-Bounty für alle Services von Google, das Belohnungen auszahlt, alleine 17 Millionen US-$ im 2025. Diese Prämie wurde über mehr als 700 Analysten verteilt, die Fehler fanden in Android, der Google Cloud, aber auch in verbreiteten und häufig gebrauchten Open Source Programmen und Bibliotheken.
-
#C4DT_EPFL #WeeklyPicks Google startete schon im 2010 ein 'Vulnerability Reward Program', ein Bug-Bounty für alle Services von Google, das Belohnungen auszahlt, alleine 17 Millionen US-$ im 2025. Diese Prämie wurde über mehr als 700 Analysten verteilt, die Fehler fanden in Android, der Google Cloud, aber auch in verbreiteten und häufig gebrauchten Open Source Programmen und Bibliotheken.
-
#C4DT_EPFL #WeeklyPicks Google startete schon im 2010 ein 'Vulnerability Reward Program', ein Bug-Bounty für alle Services von Google, das Belohnungen auszahlt, alleine 17 Millionen US-$ im 2025. Diese Prämie wurde über mehr als 700 Analysten verteilt, die Fehler fanden in Android, der Google Cloud, aber auch in verbreiteten und häufig gebrauchten Open Source Programmen und Bibliotheken.
-
#C4DT_EPFL #WeeklyPicks Google startete schon im 2010 ein 'Vulnerability Reward Program', ein Bug-Bounty für alle Services von Google, das Belohnungen auszahlt, alleine 17 Millionen US-$ im 2025. Diese Prämie wurde über mehr als 700 Analysten verteilt, die Fehler fanden in Android, der Google Cloud, aber auch in verbreiteten und häufig gebrauchten Open Source Programmen und Bibliotheken.
-
Having a blast at the EPFL Industry Day, stuck «inside » the deepfake exhibit of the Centre for Digital Trust #c4dt_epfl
-
Having a blast at the EPFL Industry Day, stuck «inside » the deepfake exhibit of the Centre for Digital Trust #c4dt_epfl
-
Having a blast at the EPFL Industry Day, stuck «inside » the deepfake exhibit of the Centre for Digital Trust #c4dt_epfl
-
Having a blast at the EPFL Industry Day, stuck «inside » the deepfake exhibit of the Centre for Digital Trust #c4dt_epfl
-
#WeeklyPick #C4DT_EPFL #EPFL #digiges #SocNum @digiges
Un sujet qui m'inquiète en tant que supporteur des logiciels libres et de la souveraineté qui les accompagne: Google veut fermeture sa plate-forme Android pour toute application qui n'est pas signé par un développeur agréé par Google. Le sujet fait débat surtout en Europe, où Apple a dû ouvrir ses iPhones aux applications étrangères.
-
#WeeklyPick #C4DT_EPFL #EPFL #digiges #SocNum @digiges
Un sujet qui m'inquiète en tant que supporteur des logiciels libres et de la souveraineté qui les accompagne: Google veut fermeture sa plate-forme Android pour toute application qui n'est pas signé par un développeur agréé par Google. Le sujet fait débat surtout en Europe, où Apple a dû ouvrir ses iPhones aux applications étrangères.
-
#WeeklyPick #C4DT_EPFL #EPFL #digiges #SocNum @digiges
Un sujet qui m'inquiète en tant que supporteur des logiciels libres et de la souveraineté qui les accompagne: Google veut fermeture sa plate-forme Android pour toute application qui n'est pas signé par un développeur agréé par Google. Le sujet fait débat surtout en Europe, où Apple a dû ouvrir ses iPhones aux applications étrangères.
-
#WeeklyPick #C4DT_EPFL #EPFL #digiges #SocNum @digiges
Un sujet qui m'inquiète en tant que supporteur des logiciels libres et de la souveraineté qui les accompagne: Google veut fermeture sa plate-forme Android pour toute application qui n'est pas signé par un développeur agréé par Google. Le sujet fait débat surtout en Europe, où Apple a dû ouvrir ses iPhones aux applications étrangères.
-
I always liked Anthropic's stance of doing the good thing and setting standards with regards to security and ehtics in LLMs. Unfortunately they changed their mind and will now only define 'goals that we will openly grade our progress towards'. So these goals will be non-binding. At least they kept the rules of 'no AI controlled weapons' and 'no mass domestic surveillance'.
https://edition.cnn.com/2026/02/25/tech/anthropic-safety-policy-change
-
I always liked Anthropic's stance of doing the good thing and setting standards with regards to security and ehtics in LLMs. Unfortunately they changed their mind and will now only define 'goals that we will openly grade our progress towards'. So these goals will be non-binding. At least they kept the rules of 'no AI controlled weapons' and 'no mass domestic surveillance'.
https://edition.cnn.com/2026/02/25/tech/anthropic-safety-policy-change
-
I always liked Anthropic's stance of doing the good thing and setting standards with regards to security and ehtics in LLMs. Unfortunately they changed their mind and will now only define 'goals that we will openly grade our progress towards'. So these goals will be non-binding. At least they kept the rules of 'no AI controlled weapons' and 'no mass domestic surveillance'.
https://edition.cnn.com/2026/02/25/tech/anthropic-safety-policy-change
-
I always liked Anthropic's stance of doing the good thing and setting standards with regards to security and ehtics in LLMs. Unfortunately they changed their mind and will now only define 'goals that we will openly grade our progress towards'. So these goals will be non-binding. At least they kept the rules of 'no AI controlled weapons' and 'no mass domestic surveillance'.
https://edition.cnn.com/2026/02/25/tech/anthropic-safety-policy-change
-
We are pleased to share some output of our work on electronic credentials in our technical blog: "Choosing a Cryptographic Library for Anonymous Credentials":
https://eid-privacy.github.io/wp2/2026/01/27/docknetwork-crypto-library.html
-
We are pleased to share some output of our work on electronic credentials in our technical blog: "Choosing a Cryptographic Library for Anonymous Credentials":
https://eid-privacy.github.io/wp2/2026/01/27/docknetwork-crypto-library.html
-
We are pleased to share some output of our work on electronic credentials in our technical blog: "Choosing a Cryptographic Library for Anonymous Credentials":
https://eid-privacy.github.io/wp2/2026/01/27/docknetwork-crypto-library.html
-
We are pleased to share some output of our work on electronic credentials in our technical blog: "Choosing a Cryptographic Library for Anonymous Credentials":
https://eid-privacy.github.io/wp2/2026/01/27/docknetwork-crypto-library.html
-
Long live Friday: here is a short article pointing to our Proof-of-Concept on privacy-preserving #eID presentation of credentials. It includes proofs for issuer signature, holder binding, predicate proofs, and non-revocation proofs.
Spoiler alert: #NoirLang is awesome! It's easy to understand AND fast!
Have a look here:
-
Long live Friday: here is a short article pointing to our Proof-of-Concept on privacy-preserving #eID presentation of credentials. It includes proofs for issuer signature, holder binding, predicate proofs, and non-revocation proofs.
Spoiler alert: #NoirLang is awesome! It's easy to understand AND fast!
Have a look here:
-
Long live Friday: here is a short article pointing to our Proof-of-Concept on privacy-preserving #eID presentation of credentials. It includes proofs for issuer signature, holder binding, predicate proofs, and non-revocation proofs.
Spoiler alert: #NoirLang is awesome! It's easy to understand AND fast!
Have a look here:
-
Long live Friday: here is a short article pointing to our Proof-of-Concept on privacy-preserving #eID presentation of credentials. It includes proofs for issuer signature, holder binding, predicate proofs, and non-revocation proofs.
Spoiler alert: #NoirLang is awesome! It's easy to understand AND fast!
Have a look here:
-
Finland is combatting Fake News at its destination. It teaches media literacy in schools, starting from the age of three, on through the whole curriculum of the students. This approach is paying off, with the country ranking at the top of the European Media Literacy Index. Meanwhile, here in Switzerland, students are mostly left to figure out for themselves what is real and what is fake.
-
Finland is combatting Fake News at its destination. It teaches media literacy in schools, starting from the age of three, on through the whole curriculum of the students. This approach is paying off, with the country ranking at the top of the European Media Literacy Index. Meanwhile, here in Switzerland, students are mostly left to figure out for themselves what is real and what is fake.
-
Finland is combatting Fake News at its destination. It teaches media literacy in schools, starting from the age of three, on through the whole curriculum of the students. This approach is paying off, with the country ranking at the top of the European Media Literacy Index. Meanwhile, here in Switzerland, students are mostly left to figure out for themselves what is real and what is fake.
-
Finland is combatting Fake News at its destination. It teaches media literacy in schools, starting from the age of three, on through the whole curriculum of the students. This approach is paying off, with the country ranking at the top of the European Media Literacy Index. Meanwhile, here in Switzerland, students are mostly left to figure out for themselves what is real and what is fake.
-
EPFL and CYD Campus announce the 13th call for Cyber-Defence Fellowships. Opportunities include: Doctoral Fellowships (up to 4 years), Distinguished Postdoctoral Fellowships (up to 2 years) – both with deadline Feb 25, 2026; Master Thesis Fellowships (6 months) and Proof of Concept Fellowships (12 months) – rolling calls. Online applicant workshop: Jan 13, 2026. Supported by armasuisse Science and Technology.
Contact: [email protected]
https://actu.epfl.ch/news/13th-call-for-proposals-cyber-defence-cyd-fellow-2/
-
EPFL and CYD Campus announce the 13th call for Cyber-Defence Fellowships. Opportunities include: Doctoral Fellowships (up to 4 years), Distinguished Postdoctoral Fellowships (up to 2 years) – both with deadline Feb 25, 2026; Master Thesis Fellowships (6 months) and Proof of Concept Fellowships (12 months) – rolling calls. Online applicant workshop: Jan 13, 2026. Supported by armasuisse Science and Technology.
Contact: [email protected]
https://actu.epfl.ch/news/13th-call-for-proposals-cyber-defence-cyd-fellow-2/
-
EPFL and CYD Campus announce the 13th call for Cyber-Defence Fellowships. Opportunities include: Doctoral Fellowships (up to 4 years), Distinguished Postdoctoral Fellowships (up to 2 years) – both with deadline Feb 25, 2026; Master Thesis Fellowships (6 months) and Proof of Concept Fellowships (12 months) – rolling calls. Online applicant workshop: Jan 13, 2026. Supported by armasuisse Science and Technology.
Contact: [email protected]
https://actu.epfl.ch/news/13th-call-for-proposals-cyber-defence-cyd-fellow-2/
-
EPFL and CYD Campus announce the 13th call for Cyber-Defence Fellowships. Opportunities include: Doctoral Fellowships (up to 4 years), Distinguished Postdoctoral Fellowships (up to 2 years) – both with deadline Feb 25, 2026; Master Thesis Fellowships (6 months) and Proof of Concept Fellowships (12 months) – rolling calls. Online applicant workshop: Jan 13, 2026. Supported by armasuisse Science and Technology.
Contact: [email protected]
https://actu.epfl.ch/news/13th-call-for-proposals-cyber-defence-cyd-fellow-2/
-
Why do we measure LLM agents? What do we measure? What are the normal values?
- Julia Bazinska, Senior Research Engineer, Lakera AI
- Prof. Martin Jaggi, Head of Machine Learning and Optimization Laboratory, EPFL
- Mark Müller, Co-Founder and CTO, LogicStar AI
- Lea Strohm, Senior Technical Product Manager, Thomson ReutersModerated by Imad Aad from C4DT (who is a bit nervous :)
-
Why do we measure LLM agents? What do we measure? What are the normal values?
- Julia Bazinska, Senior Research Engineer, Lakera AI
- Prof. Martin Jaggi, Head of Machine Learning and Optimization Laboratory, EPFL
- Mark Müller, Co-Founder and CTO, LogicStar AI
- Lea Strohm, Senior Technical Product Manager, Thomson ReutersModerated by Imad Aad from C4DT (who is a bit nervous :)
-
Why do we measure LLM agents? What do we measure? What are the normal values?
- Julia Bazinska, Senior Research Engineer, Lakera AI
- Prof. Martin Jaggi, Head of Machine Learning and Optimization Laboratory, EPFL
- Mark Müller, Co-Founder and CTO, LogicStar AI
- Lea Strohm, Senior Technical Product Manager, Thomson ReutersModerated by Imad Aad from C4DT (who is a bit nervous :)
-
Why do we measure LLM agents? What do we measure? What are the normal values?
- Julia Bazinska, Senior Research Engineer, Lakera AI
- Prof. Martin Jaggi, Head of Machine Learning and Optimization Laboratory, EPFL
- Mark Müller, Co-Founder and CTO, LogicStar AI
- Lea Strohm, Senior Technical Product Manager, Thomson ReutersModerated by Imad Aad from C4DT (who is a bit nervous :)
-
Lea Strohm from Thomson Reuters on "Building Trust in Agentic Systems".
Interesting high-level thoughts about LLM agents and things that is not a big problem for programmers, but not good for others, which are not used to "my computer is only doing what I ask it and it's normal for it to fail". This includes lawyers, but also policy makers, and of course news agencies which want to make sure they have correct information.
It's "TR" - never heard that abbreviation before...
-
Lea Strohm from Thomson Reuters on "Building Trust in Agentic Systems".
Interesting high-level thoughts about LLM agents and things that is not a big problem for programmers, but not good for others, which are not used to "my computer is only doing what I ask it and it's normal for it to fail". This includes lawyers, but also policy makers, and of course news agencies which want to make sure they have correct information.
It's "TR" - never heard that abbreviation before...
-
Lea Strohm from Thomson Reuters on "Building Trust in Agentic Systems".
Interesting high-level thoughts about LLM agents and things that is not a big problem for programmers, but not good for others, which are not used to "my computer is only doing what I ask it and it's normal for it to fail". This includes lawyers, but also policy makers, and of course news agencies which want to make sure they have correct information.
It's "TR" - never heard that abbreviation before...
-
Lea Strohm from Thomson Reuters on "Building Trust in Agentic Systems".
Interesting high-level thoughts about LLM agents and things that is not a big problem for programmers, but not good for others, which are not used to "my computer is only doing what I ask it and it's normal for it to fail". This includes lawyers, but also policy makers, and of course news agencies which want to make sure they have correct information.
It's "TR" - never heard that abbreviation before...
-
Mark Müller is up next, from LogicStar AI. He talks about "Code Agents in the Real World". Here is his take on the definition of an "AI Agent".
Quite close to my favorite: "An AI Agent is an LLM in a loop interacting with tools"...
-
Mark Müller is up next, from LogicStar AI. He talks about "Code Agents in the Real World". Here is his take on the definition of an "AI Agent".
Quite close to my favorite: "An AI Agent is an LLM in a loop interacting with tools"...
-
Mark Müller is up next, from LogicStar AI. He talks about "Code Agents in the Real World". Here is his take on the definition of an "AI Agent".
Quite close to my favorite: "An AI Agent is an LLM in a loop interacting with tools"...
-
Mark Müller is up next, from LogicStar AI. He talks about "Code Agents in the Real World". Here is his take on the definition of an "AI Agent".
Quite close to my favorite: "An AI Agent is an LLM in a loop interacting with tools"...
-
Julia Bazinska from Lakera (https://gandalf.lakera.ai/baseline) is up next.
AI Chatbots increase the attack surface, which give a bunch a different issues. So far it's mostly "screenshot attacks", where bad outputs of the bots are shown.
Now AI Agents increase the attack surface even more. And can do actual harm to your environment.
It's all the fault of the MCPs :) https://de.wikipedia.org/wiki/Model_Context_Protocol
-
Julia Bazinska from Lakera (https://gandalf.lakera.ai/baseline) is up next.
AI Chatbots increase the attack surface, which give a bunch a different issues. So far it's mostly "screenshot attacks", where bad outputs of the bots are shown.
Now AI Agents increase the attack surface even more. And can do actual harm to your environment.
It's all the fault of the MCPs :) https://de.wikipedia.org/wiki/Model_Context_Protocol
-
Julia Bazinska from Lakera (https://gandalf.lakera.ai/baseline) is up next.
AI Chatbots increase the attack surface, which give a bunch a different issues. So far it's mostly "screenshot attacks", where bad outputs of the bots are shown.
Now AI Agents increase the attack surface even more. And can do actual harm to your environment.
It's all the fault of the MCPs :) https://de.wikipedia.org/wiki/Model_Context_Protocol
-
Julia Bazinska from Lakera (https://gandalf.lakera.ai/baseline) is up next.
AI Chatbots increase the attack surface, which give a bunch a different issues. So far it's mostly "screenshot attacks", where bad outputs of the bots are shown.
Now AI Agents increase the attack surface even more. And can do actual harm to your environment.
It's all the fault of the MCPs :) https://de.wikipedia.org/wiki/Model_Context_Protocol