AI Research
A language model built for the public good
Earlier this week in Geneva, around 50 leading global initiatives and organisations dedicated to open-source LLMs and trustworthy AI convened at the International Open-Source LLM Builders Summit. Hosted by the AI centres of EPFL and ETH Zurich, the event marked a significant step in building a vibrant and collaborative international ecosystem for open foundation models. Open LLMs are increasingly viewed as credible alternatives to commercial systems, most of which are developed behind closed doors in the United States or China.
Participants of the summit previewed the forthcoming release of a fully open, publicly developed LLM — co-created by researchers at EPFL, ETH Zurich and other Swiss universities in close collaboration with engineers at CSCS. Currently in final testing, the model will be downloadable under an open license. The model focuses on transparency, multilingual performance, and broad accessibility.
The model will be fully open: source code and weights will be publicly available, and the training data will be transparent and reproducible, supporting adoption across science, government, education, and the private sector. This approach is designed to foster both innovation and accountability.
“Fully open models enable high-trust applications and are necessary for advancing research about the risks and opportunities of AI. Transparent processes also enable regulatory compliance,” says Imanol Schlag, research scientist at the ETH AI Center, who is leading the effort alongside EPFL AI Center faculty members and professors Antoine Bosselut and Martin Jaggi.
Multilingual by design
A defining characteristic of the LLM is its fluency in over 1000 languages. “We have emphasised making the models massively multilingual from the start,” says Antoine Bosselut.
Training of the base model was done on a large text dataset in over 1500 languages — approximately 60% English and 40% non-English languages — as well as code and mathematics data. Given the representation of content from all languages and cultures, the resulting model maintains the highest global applicability.
Designed for scale and inclusion
The model will be released in two sizes — 8 billion and 70 billion parameters, meeting a broad range of users’ needs. The 70B version will rank among the most powerful fully open models worldwide. The number of parameters reflects a model’s capacity to learn and generate complex responses.
High reliability is achieved through training on over 15 trillion high-quality training tokens (units representing a word or part of the word), enabling robust language understanding and versatile use cases.
Responsible data practices
The LLM is being developed with due consideration to Swiss data protection laws, Swiss copyright laws, and the transparency obligations under the EU AI Act. In a external page recent study, the project leaders demonstrated that for most everyday tasks and general knowledge acquisition, respecting web crawling opt-outs during data acquisition produces virtually no performance degradation.
Supercomputer as an enabler of sovereign AI
The model is trained on the “Alps” supercomputer at CSCS in Lugano, one of the world’s most advanced AI platforms, equipped with over 10,000 NVIDIA Grace Hopper Superchips. The system’s scale and architecture made it possible to train the model efficiently using 100% carbon-neutral electricity.
The successful realisation of “Alps” was significantly facilitated by a long-standing collaboration spanning over 15 years with NVDIA and HPE/Cray. This partnership has been pivotal in shaping the capabilities of “Alps”, ensuring it meets the demanding requirements of large-scale AI workloads, including the pre-training of complex LLMs.
“Training this model is only possible because of our strategic investment in ‘Alps’, a supercomputer purpose-built for AI,” says Thomas Schulthess, Director of CSCS and professor at ETH Zurich. “Our enduring collaboration with NVIDIA and HPE exemplifies how joint efforts between public research institutions and industry leaders can drive sovereign infrastructure, fostering open innovation — not just for Switzerland, but for science and society worldwide.”
Public access and global reuse
In late summer, the LLM will be released under the Apache 2.0 License. Accompanying documentation will detail the model architecture, training methods, and usage guidelines to enable transparent reuse and further development.
“As scientists from public institutions, we aim to advance open models and enable organiations to build on them for their own applications”, says Antoine Bosselut.
“By embracing full openness — unlike commercial models that are developed behind closed doors — we hope that our approach will drive innovation in Switzerland, across Europe, and through multinational collaborations. Furthermore, it is a key factor in attracting and nurturing top talent,” says EPFL professor Martin Jaggi.
AI Research
Digital Agency Fuel Online Launches AI SEO Research Division,
Boston, MA – As Google continues to reshape the digital landscape with its Search Generative Experience (SGE) and AI-powered search results, Fuel Online [https://fuelonline.com/] is blazing a trail as the nation’s leading agency in AI SEO [https://fuelonline.com/]and SGE optimization [https://fuelonline.com/].
Recognizing the urgent need for businesses to adapt to AI-first search engines, Fuel Online has launched a dedicated AI SEO Research & Development Division focused exclusively on decoding how AI models like Google SGE read, rank, and render web content. The division’s mission: to test, reverse-engineer, and deploy cutting-edge strategies that future-proof clients’ visibility in an era of AI-generated search answers.
“AI is not the future of SEO – it’s the present . If your content doesn’t rank in SGE, it may never be seen. That’s why we’re investing heavily in understanding and optimizing for how large language models surface content,” said Scott Levy, CEO of Fuel Online Digital Marketing Agency [https://fuelonline.com/].
Fuel Online’s Digital Marketing team is already helping Fortune 500 brands, high-growth startups, and ecommerce leaders gain traction in AI-powered results using proprietary tactics including:
* NLP entity linking & semantic schema
* SGE-optimized content blocks & voice search targeting
* AI-readiness audits tailored for Google’s evolving ranking models
As detailed in their comprehensive Google SGE & AI Optimization Guide [https://fuelonline.com/insights/google-sge-and-ai-optimization-guide-how-to-optimize/], Fuel Online offers strategic insight into aligning websites with Google’s new generative layer. The agency also provides live testing environments, allowing clients to see firsthand how AI engines interpret their content. Why This Matters: According to industry data, click-through rates have dropped by up to 60% on some keywords since the rollout of SGE, as users get direct AI-generated answers instead of traditional blue links. Fuel Online’s AI SEO division helps clients reclaim that lost visibility and win placement inside AI search results. With over two decades of award-winning digital strategy under its belt and a reputation as one of the top digital marketing agencies in the U.S., Fuel Online is once again setting the standard – this time for the AI optimization era.
Media Contact:
Fuel Online
Boston, MA
(888)-475-2552
https://FuelOnline.com
Media Contact
Company Name: Fuel Online
Contact Person: Media Relation Management
Email:Send Email [https://www.abnewswire.com/email_contact_us.php?pr=digital-agency-fuel-online-launches-ai-seo-research-division-cementing-its-position-at-the-forefront-of-sge-optimization]
Phone: (888)-475-2552
City: Boston
State: MA
Country: United States
Website: https://fuelonline.com
Legal Disclaimer: Information contained on this page is provided by an independent third-party content provider. ABNewswire makes no warranties or responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you are affiliated with this article or have any complaints or copyright issues related to this article and would like it to be removed, please contact retract@swscontact.com
This release was published on openPR.
AI Research
AI, Wrong Guy: Investigating the use and dangers of artificial intelligence in Jacksonville policing – Action News Jax
AI Research
Investigating the use and dangers of artificial intelligence in Jacksonville policing
A Lee County man was wrongfully arrested last year after AI facial recognition technology used by the Jacksonville Sheriff’s Office got it wrong. Experts are now warning about the potential dangers of the technology.
The Jacksonville Beach Police Department said 51-year-old Robert Dillon allegedly tried luring a 12-year-old child in Jacksonville Beach back in November of 2023. According to a police report, Dillon was linked to a suspect caught on surveillance video in a Jacksonville Beach McDonald’s through the use of the Jacksonville Sheriff’s Office’s AI facial recognition technology.
Jacksonville Beach PD conferred with JSO, according to the report, and the technology found a 93% match between Dillon and the suspect using that technology. The report says police then provided a photo spread of Dillon and other similar-looking individuals to two witnesses. Both identified Dillon as the suspect.
However, the case would later be completely dropped. The state attorney’s office told Action News Jax the arrest will be wiped from Mr. Dillon’s record.
“Police are not allowed under the Constitution to arrest somebody without probable cause,” Nate Freed-Wessler with the American Civil Liberties Union would later tell Action News Jax. “And this technology expressly cannot provide probable cause, it is so glitchy, it’s so unreliable. At best, it has to be viewed as an extremely unreliability lead because it often, often gets it wrong.”
Freed-Wessler is the deputy director for the ACLU’s Speech, Privacy, and Technology Project. He was also part of the legal team that helped sue on behalf of Robert Williams – a Detroit man wrongfully arrested thanks to facial recognition similar to the technology used to identify Dillon. The Detroit Police department settled that case for $300,000 in damages, and implemented safeguards when using AI facial recognition in their investigations.
Freed-Wessler told Action News Jax that wrongful arrests using AI facial recognition are more common than many think, especially among people of color.
“It’s partly because of photo quality problems in low light situations, when the cameras are trying to identify darker skin people,” Freed-Wessler explained. “In fact, in almost all of the wrongful arrest cases around the country that we know of, it’s been black people who have been incorrectly, wrongfully picked up by police.”
Action News Jax sat down with Jacksonville Sheriff T.K. Waters to discuss the use of AI facial recognition technology in Jacksonville Sheriff’s Office investigations. Sheriff Waters reassured the technology is simply a small piece of the investigative puzzle.
“If you came to me with a facial recognition hit and that was your probable cause, I would probably kick you out of my office because that’s not how it works,” Sheriff Waters explained. “And I can’t speak to [the Jacksonville Beach Police Department’s] investigation. I can tell you this, there better be a lot more that goes along with that to help make sure that we have the proper individual too.”
[DOWNLOAD: Free Action News Jax app for alerts as news breaks]
However, Freed-Wessler believes this procedure wasn’t properly followed by Jacksonville Beach police in their investigation, adding that photo spreads based on a facial recognition match aren’t sufficient evidence to make an arrest.
“When this technology gets it wrong, it’s going to get it wrong with a face of somebody who looks similar to the suspect,” Freed-Wessler explained. “It’s no surprise that when police juice a lineup procedure with a doppelganger, with a lookalike, a witness is going to choose an innocent person.”
Now, the Jacksonville Beach Police Department tells Action News Jax the investigation is still open after Dillon was cleared of any wrongdoing, adding in part:
“We will not be commenting on this matter beyond stating that all warrant requests are submitted to the state attorney’s office. It is solely their decision whether or not to move forward with issuing a warrant.”
Action News Jax reached out to the state attorney’s office as well. A spokesman only confirmed Dillon was cleared of any wrongdoing.
Now, Dillon’s lawyer tells Action News Jax that he is seeking compensation, although he and Dillon declined interview requests.
Meanwhile Courtney Barclay, an AI policy expert at Jacksonville University, said law enforcement agencies across the nation will continue to use AI and facial recognition. Barclay outlined the need to always second-guess.
“Every industry is just now starting to scratch the surface of the potential of AI, how it can impact our society. Law enforcement is no exception,” Barclay said. “And so, again, we just want to be cognizant of the risks.”
>>> STREAM ACTION NEWS JAX LIVE <<<
[SIGN UP: Action News Jax Daily Headlines Newsletter]
-
Funding & Business1 week ago
Kayak and Expedia race to build AI travel agents that turn social posts into itineraries
-
Jobs & Careers1 week ago
Mumbai-based Perplexity Alternative Has 60k+ Users Without Funding
-
Mergers & Acquisitions1 week ago
Donald Trump suggests US government review subsidies to Elon Musk’s companies
-
Funding & Business1 week ago
Rethinking Venture Capital’s Talent Pipeline
-
Jobs & Careers1 week ago
Why Agentic AI Isn’t Pure Hype (And What Skeptics Aren’t Seeing Yet)
-
Education3 days ago
9 AI Ethics Scenarios (and What School Librarians Would Do)
-
Education4 days ago
Teachers see online learning as critical for workforce readiness in 2025
-
Education4 days ago
Nursery teachers to get £4,500 to work in disadvantaged areas
-
Education6 days ago
How ChatGPT is breaking higher education, explained
-
Education1 week ago
AERDF highlights the latest PreK-12 discoveries and inventions