The EPFL will launch a Swiss language model | Press review n°14
I selected important news on artificial intelligence during the week of July 7 to 13, 2025. Here's my commentary.
Welcome to the fourteenth press review of Artificial reality. This week I focused on a new Swiss langage model, the pro-Hitler messages of Grok and the use of U.S. technologies in the Middle East. Have a good read!
📰 Read
The EPFL will launch a Swiss language model
There will soon be a Swiss alternative to ChatGPT, Meta AI, Grok or Copilot. The Federal Polytechnic School of Lausanne (EPFL), a public research university in Switzerland, is developing a language model in collaboration with the Federal Institute of Technology Zurich (ETH Zurich), Le Temps reports in an article published on Wednesday.
Language models are an artificial intelligence technology on which the conversational agents such as ChatGPT are based. The chatbot of OpenAI, for example, uses different language models like GPT-4o or o3, each having its own characteristics. These models are trained on vast datasets, including texts, to learn how to respond to users’ questions in natural language.
Many advantages
The language model developed by EPFL and ETH Zurich will be freely available to the public, developers, companies and public institutions. It represents an interesting alternative to U.S. or Chinese models due to many advantages:
Contrary to Big Tech models, often designed in secret, the Swiss model will be completely open source and the parameters learned during its training will be public.
It is trained on copyright-free data only, unlike the big U.S. companies which often use lots of contents without the consent of their creators.
It will master more than 1500 languages.
It is developed with the expertise of the two technology universities’ engineers, including scientists who work in the AI Centers of these institutions.
It is trained on the supercomputer Alps of the Swiss National Supercomputing Center (CSCS) in Lugano. It is the most powerful computer in Europe and one of the most advanced academic AI platforms in the world.
It will be possible to download the model to run it locally on a computer.
Partners of the two technology universities will develop an interface similar to ChatGPT to allow the public to easily use the model from the beginning of September.
The language model of EPFL and ETH Zurich will be launched at the end of the summer with two sizes, 8 billion and 70 billion parameters. According to the technology universities, the 70 billion parameters version should be one of the most powerful fully-open source models in the world.
“We estimate that our model is at the level of the commercial models available on the market. And we already plan to update it on a regular basis to stay in the race,” asserts Martin Jaggin of the AI Center of EPFL.
Watch (in French): Intelligence artificielle: un modèle de langage suisse (RTS) and Présentation du modèle de langage suisse d’IA avec Martin Rajman (RTS)
Seven important news this week
Trump’s Big Beautiful Gift to Anduril (The Intercept)
Grok Is the Latest in a Long Line of Chatbots to Go Full Nazi (The Intercept)
A.I.-Generated Images of Child Sexual Abuse Are Flooding the Internet (The New York Times)
Google's AI Overviews hit by EU antitrust complaint from independent publishers (Reuters)
Inside OpenAI’s empire: A conversation with Karen Hao (MIT Technology Review)
The AI Scraping Fight That Could Change the Future of the Web (The Wall Street Journal)
Meta Invests $3.5 Billion in World’s Largest Eye-Wear Maker in AI Glasses Push (Bloomberg)
Read the other articles of the week I have selected by clicking here.
🎥 Watch
Grok says he is “MechaHitler”
The chatbot of xAI, the artificial intelligence company of Elon Musk, made remarks favorable to Adolf Hitler on X after it was updated.
The political commentator Kyle Kulinski talked about this incident and shared many examples of pro-Nazi and antisemitic messages that Grok posted during the week.
‘CALL ME MECHAHlTLER’: Grok Goes FULL NAZl After Elon Update
Seven important videos this week
Silicon Valley Wants Endless War: How Venture Capital Fuels Racist Killing (Taylor Lorenz)
Apple Loses Top AI Exec to Meta’s Hiring Spree (Bloomberg)
The US, China Race for AI Supremacy (Bloomberg)
Microsoft says its AI tops doctors in medical diagnoses (Yahoo)
Why Nvidia Will Still Lead AI for the Next Decade (Bloomberg)
Watch the other videos of the week I have selected by clicking here.
🔈 Listen
The technological empire of the United States in the Middle East
In the new episode of Tech Won’t Save Us, journalist Paris Marx discusses the use of U.S. technologies in the Middle East with Laleh Khalili, Professor of Gulf Studies at the University of Exeter, in England, and writer of Extractive Capitalism.
“That marriage between the U.S. war-making capabilities and its technological advancement has actually been inseparable from the very beginning of the United States’ birth.”
- Laleh Khalili
Thank you for reading the fourteenth press review of Artificial reality! Subscribe for free to receive the reviews directly into your mailbox. With a paid subscription, you will also have access to all articles and to an exclusive monthly newsletter.
Have a good week,
Arnaud
