👏 Congratulations on this achievement and all the best for Cecilia’s new role as postdoctoral researcher at the @cam.ac.uk!

#NLP #PhDDefense #MultilingualAI #CulturalAI #LanguageModels #UKPLab #TUDarmstadt @cs-tudarmstadt.bsky.social

Cohere's Tiny Aya Fits 70 Languages Into 3.35 Billion Parameters and Runs on a Phone Cohere Labs releases Tiny Aya, a 3.35B open-weight multilingual model that beats Gemma 3 4B in 46 of 61 languages on translation and runs at 32 tokens/sec on an iPhone.

Cohere's Tiny Aya Fits 70 Languages Into 3.35 Billion Parameters and Runs on a Phone

awesomeagents.ai/news/cohere-tiny-aya-mul...

#Cohere #TinyAya #MultilingualAi

Cohere’s Tiny Aya: The 3.35B-Parameter Model Bringing 70+ Languages to Your Pocket Most AI models demand English fluency, stable broadband, and a cloud server to function. Cohere’s Tiny Aya removes all three requirements at once. Launched February 17, 2026, at the India AI Summit, this open-

Cohere's Tiny Aya runs 70+ languages offline even on a phone. 3.35B parameters, 5 regional variants, and 32 tokens/sec on iPhone 17 Pro. AdwaitX breaks down everything confirmed at launch. #AdwaitX #TinyAya #MultilingualAI

Here’s a BizGreetz AI® demo for a concept store in Paris — created in minutes and highlighting support for 30+ languages.

One platform. Global creative.

Unmute, and watch 🎬

#CreativeAI #VideoMarketing #MultilingualAI #BizGreetz #AdGreetz

Endpoints 2025 - Lora Aroyo: Multimodal Safety: Assessing Global and Cultural Sensitivities in AI YouTube video by MLCommons

Dr. Lora Aroyo (@GoogleDeepMind) on building benchmarks that actually work across languages, cultures, and modalities.

Most benchmarks are too narrow. This is how we fix that.

Watch: https://bit.ly/4bAomff

#Endpoints2025 #MultilingualAI

🎙️ In this interview, Javier García Gilabert and Miguel Claramunt from Barcelona Supercomputing Center share their perspective on the journey so far: eloquenceai.eu/partner-pers...

#ELOQUENCEAI #TrustworthyAI #MultilingualAI #ConversationalAI #HorizonEurope #AIResearch

AI Contact Center: Multilingual Intelligence Transforming Scalable Enterprise Support | business IntroductionWhen it comes to outsourcing customer service for enterprises, a ...

#AIContactCenter #MultilingualAI #CustomerSupport #EnterpriseCX

📣Today, we have the honor of announcing our second speaker of the ELOQUENCE Webcafé – Lorenzo Concina.

🗓️ Date: December 18 2025
🕐 Time: 11:00 AM (CET)

🔗 Registration is required – save your spot now: us02web.zoom.us/meeting/regi...

#ELOQUENCEAI #Webcafé #HorizonEurope #MultilingualAI

CAMB AI has partnered with Kompact AI by Ziroh Labs, to optimize multilingual voice models for CPUs.

#CambAI #KompactAI #ZirohLabs #AIPartnership #MultilingualAI #VoiceAI #deccanfounders

A pop-up message in Google Sheets shows a blue sparkle icon and the text “Gemini editing requests are only available in English at this time,” with options below for Insert, Copy, a menu button, thumbs up, and thumbs down.

Gemini in Google Sheets doesn’t speak Croatian.
It barely speaks British English.

Still manually correcting “prioritize” → “prioritise” like it’s 1999.
Global AI, local problems.

#GoogleSheets #Gemini #AI #Croatian #BritishEnglish #LanguageMatters #MultilingualAI #LLM #FutureOfWork #GoogleAI

Sponsorship opportunities for #EAMT2026 🎉
15-18 June, Tilburg 🇳🇱
Support #MT & #MultilingualAI research + community building.
Themed options:
🌱 Sustainability
🤟 Accessibility / Sign-language
♻️ Eco-swag
Interested in sponsoring?
More Info at eamt2026@tilburguniversity.edu

Model selection is crucial for RAG, particularly for non-English languages. Performance varies greatly, so choose appropriate models and consult resources for suitable multilingual embedding models to ensure broad language support. #MultilingualAI 5/6

📣 Meet our second guest speaker for the upcoming ELOQUENCE webinar on “Trustworthy AI in High-Stakes Applications” – Dr. Petr Motlicek!

🔗Link for registration: us02web.zoom.us/meeting/regi...

#ELOQUENCEAI #TrustLLM #TrustworthyAI #HighStakesAI #Webinar #HorizonEurope #MultilingualAI #AI

Multilingual Isn’t Cross-Lingual: Inside My Benchmark of 11 LLMs on Mid- & Low-Resource Languages

A data-driven look at why multilingual LLMs fail at true cross-lingual reasoning, and how culturally intact datasets and new metrics reveal the gap. #multilingualai

🌟 Join us for the next ELOQUENCE x TrustLLM
Webinar! 🌟

🗓️ Date: November 27th, 2025
🕐 Time: 13:00 CET

📌 Save your spot: us02web.zoom.us/meeting/regi...

#ELOQUENCEAI #TrustLLM #TrustworthyAI #AI #HighStakesAI #Webinar #HorizonEurope #MultilingualAI

🌍 Friday AI Fact
Did you know that grammatical gender can shape how LLMs express gender bias? An ELOQUENCE study shows big differences across languages.
📖Read more here: eloquenceai.eu/publications
#ELOQUENCEAI #GenderBias #MultilingualAI

AI4Bharat just launched the Indic LLM Arena – a new benchmark playground for Indian language models, handling code‑switching, multilingual nuance, and safety. Curious how it’ll shape the IndiaAI Mission? Dive in! #AI4Bharat #IndicLLMArena #MultilingualAI

🔗 aidailypost.com/news/ai4bhar...

One year ago, we launched the first season of the #ELOQUENCE podcast that explores the fascinating crossroads of large language models and ethical considerations in AI.

🎧 Tune in and explore the full playlist on our YouTube channel: www.youtube.com/watch?v=6J67...

#LLM #EthicalAI #MultilingualAI

✨ELOQUENCE at Interspeech 2025
At this year’s Interspeech, ELOQUENCE showcased cutting-edge research advancing multilingual, trustworthy, and inclusive AI.
Read more in our latest blog 👉 eloquenceai.eu/eloquence-at...
#ELOQUENCEAI #Interspeech2025 #SpeechTechnology #TrustworthyAI #MultilingualAI

Polish emerges as top language in multilingual AI benchmark testing Researchers test 26 languages across seven synthetic tasks at context lengths from 8K to 128K tokens, finding performance gaps widen at longer contexts for low-resource languages.

Polish emerges as top language in multilingual AI benchmark testing #AI #MachineLearning #PolishLanguage #MultilingualAI #BenchmarkTesting

Polish emerges as top language in multilingual AI benchmark testing Researchers test 26 languages across seven synthetic tasks at context lengths from 8K to 128K tokens, finding performance gaps widen at longer contexts for low-resource languages.

Polish emerges as top language in multilingual AI benchmark testing #AI #MachineLearning #PolishLanguage #MultilingualAI #BenchmarkTesting

Adobe Firefly Launches New Feature for Multilingual AI Voiceovers Adobe’s new feature lets users create AI voiceovers in more than 20 languages with adjustable pacing and emotion.

Adobe Firefly Introduces Multilingual AI Voiceovers | Slator

Adobe Firefly takes creativity to the next level with its new multilingual AI voiceover feature.

Read More : slator.com/adobe-firefl...

#AdobeFirefly #AIVoiceover #MultilingualAI #AICreativity #VoiceTechnology #Localization

Linguistic diversity is a key challenge for EuroLLM. HN highlighted the difficulty of accurately representing languages like Maltese or Frisian, emphasizing the critical role of cultural context for effective multilingual models. #MultilingualAI 2/6

Sometimes the best tech solutions start with analog thinking - notebooks, post-its, and a clear desk.

Building at the intersection of #AI + #Languages + #Data

#MachineLearning #LangChain #EdTech #MultilingualAI #DataScience #AIEducation #LanguageLearning #TechLife

AI Tools That Help You Write in 10+ Languages - Article For U Unlock global voices with AI tools that craft flawless content in over 10 languages. From swapping slang to perfect grammar, these smart assistants break barriers and let your ideas travel the world effortlessly.

AI Tools That Help You Write in 10+ Languages #AI #AIWritingTools #ArtificialIntelligence #contentcreation #GlobalCommunication #languagelearning #LanguageTechnology #LanguageTools #languagetranslation #MachineLearning #MultilingualAI #MultilingualWriting #naturallanguageprocessing #ProductivityTool

MulVuln introduces multilingual AI model for vulnerability detection

MulVuln, a multilingual AI model, was tested on the REEF dataset with 4,466 CVEs and 30,987 patches across seven languages, beating baselines by up to 23.59% in F1 score. Read more: getnews.me/mulvuln-introduces-multi... #mulvuln #multilingualai

Cross‑Lingual Consistency in Multilingual Language Models Explored

At EMNLP 2025 researchers showed that code‑switching fine‑tuning improves multilingual models' cross‑lingual consistency, and activation‑patching further cuts mismatches. Read more: getnews.me/cross-lingual-consistenc... #multilingualai #crosslingual

🍂Kicking Off Autumn with Fresh Updates

📬 Check out our latest ELOQUENCE newsletter!

🔗Read the full newsletter at: eloquenceai.eu/publications/

#ELOQUENCEAI #Newsletter #LLMs #TrustworthyAI #MultilingualAI

Reinforcement Learning Improves Cross‑Lingual Reasoning in LLMs

RL fine‑tuning of Qwen2.5‑3B‑Base beats supervised tuning on math, commonsense and scientific reasoning, delivering higher accuracy and cross‑lingual transfer. Read more: getnews.me/reinforcement-learning-i... #reinforcementlearning #multilingualai #llm

Evaluating Cultural Nuance in Multilingual AI Translation

A pilot study of 87 AI translations across 24 dialects in 20 languages found grammatically correct output but frequent literal rendering of idioms and puns, missing cultural nuance. Read more: getnews.me/evaluating-cultural-nuan... #multilingualai #localization