iiinova — where your digital self keeps evolving

Reposted by Lucie-Aimée Kaffee

Shayne Longpre @shaynelongpre.bsky.social · 09/12/2025

Excited to publish this piece!

Reposted by Lucie-Aimée Kaffee

Tech Policy Press @techpolicypress.bsky.social · 09/12/2025

Policymakers must recognize the open source AI ecosystem is where influence is being negotiated: not just which models exist, but which are used; not just who can train a trillion-parameter network, but who can make it deployable, modifiable, and relevant, say Lucie-Aimée Kaffee and Shayne Longpre.

techpolicy.press

Policymakers Overlook How Open Source AI Is Reshaping Global Power | TechPolicy.Press

To understand and shape the distribution of power in AI, look to the open source ecosystem, say Lucie-Aimée Kaffee and Shayne Longpre.

074

Reposted by Lucie-Aimée Kaffee

Shayne Longpre @shaynelongpre.bsky.social · 26/11/2025

Who is winning the open AI race? Our new study Economies of Open Intelligence maps @hf.co 851k models' downloads 2020→2025. 1) Power rebalance: US tech ↓; China + community ↑ 2) Models size & efficient ↑ (MoE, quant, multimodal) 3) Intermediary layers ↑ (adapters/quantizers) 4) Transparency ↓ /🧵

293

Reposted by Lucie-Aimée Kaffee

Margaret Mitchell @mmitchell.bsky.social · 29/10/2025

🤖 Did you know your voice might be cloned without your consent from just *one sentence* of audio? That's not great. So with @frimelle.bsky.social, we brainstormed a new idea for developers who want to curb malicious use: ✨The Voice Consent Gate.✨ Details, code, here: huggingface.co/blog/voice-c...

Ornate line drawing of a fence and gate, with fleur de lis tips. The gate says CONSENT where the family name usually is.

1375

Lucie-Aimée Kaffee @frimelle.bsky.social · 29/10/2025

What does it mean if anyone’s voice can be cloned and made to say whatever someone else wants? Together with @mmitchell.bsky.social we built a first prototype, a Voice Consent Gate, to explore how consent could be built into AI voice cloning itself.

110

Reposted by Lucie-Aimée Kaffee

Giada Pistilli @giadapistilli.com · 15/10/2025

What if your most personal chat logs became the next source of ad data? @frimelle.bsky.social and I wrote an op-ed for @techpolicypress.bsky.social We look at what happens when generative AI conversations (the ones we treat as private) are turned into raw material for targeted advertising.

techpolicy.press

Before AI Exploits Our Chats, Let’s Learn from Social Media Mistakes | TechPolicy.Press

Privacy in the age of conversational AI is a governance choice, write Hugging Face's Lucie-Aimé Kaffee and Giada Pistilli.

063

Reposted by Lucie-Aimée Kaffee

Tech Policy Press @techpolicypress.bsky.social · 13/10/2025

Neither in the United States nor in the European Union are regulations yet fully prepared for the mix of intimacy and monetization that AI chatbots can introduce, write Hugging Face's Lucie-Aimée Kaffee and Giada Pistilli. We need to learn the lessons from past failures on social media, they say.

techpolicy.press

Before AI Exploits Our Chats, Let’s Learn from Social Media Mistakes | TechPolicy.Press

Privacy in the age of conversational AI is a governance choice, write Hugging Face's Lucie-Aimé Kaffee and Giada Pistilli.

196

Reposted by Lucie-Aimée Kaffee

Weizenbaum-Institut @weizenbauminstitut.bsky.social · 09/10/2025

Wie kann KI offen & verantwortungsvoll gestaltet werden? Zwei neue Publikationen fassen Ergebnisse der Konferenz „Yes, we are open!?“ zusammen. Die Veröffentlichungen bieten Empfehlungen für Politik & Praxis – für eine faire, zukunftsfähige KI. 🌍🤖 www.weizenbaum-institut.de/news/detail/...

weizenbaum-institut.de

Wege zu fairer und offener KI-Governance

Welche Rahmenbedingungen sind für die verantwortliche Gestaltung von Künstlicher Intelligenz und Open Science notwendig? Zwei neue Veröffentlichungen bündeln Konferenzergebnisse und entwickeln Handlun...

286

Lucie-Aimée Kaffee @frimelle.bsky.social · 01/09/2025

Together with @giadapistilli.com we wrote “Advertisement, Privacy, and Intimacy: Lessons from Social Media for Conversational AI”. We explore the risks when ads meet chatbots & intimacy- and why open source offers a better path. huggingface.co/blog/giadap/...

huggingface.co

Advertisement, Privacy, and Intimacy: Lessons from Social Media for Conversational AI

A Blog post by Giada Pistilli on Hugging Face

020

Reposted by Lucie-Aimée Kaffee

Giada Pistilli @giadapistilli.com · 25/08/2025

🚨 Releasing INTIMA (Interactions and Machine Attachment Benchmark): an evaluation framework for measuring how AI systems handle companionship-seeking behaviors. huggingface.co/papers/2508.... Thread on what we discovered, together with @frimelle.bsky.social and @yjernite.bsky.social

huggingface.co

Paper page - INTIMA: A Benchmark for Human-AI Companionship Behavior

Join the discussion on this paper page

153

Lucie-Aimée Kaffee @frimelle.bsky.social · 21/08/2025

🤖💬 How do different AI models handle companionship? Some say GPT-5 feels “colder” than o4 - but what does that really mean when users look for emotional support? We built the AI Companionship Leaderboard to find out 👉 huggingface.co/spaces/frime...

huggingface.co

Companionship Leaderboard - a Hugging Face Space by frimelle

Browse and analyze benchmark data for different language models. View metrics like Average, Assistant Traits, and more. Easily select and display specific columns for detailed insights.

010

Lucie-Aimée Kaffee @frimelle.bsky.social · 20/08/2025

🗺️ New blog post: Old Maps, New Terrain: Updating Labour Taxonomies for the AI Era For decades, labour taxonomies like O*NET helped us understand how tech changes work. But they were built before most work became digital-first, and long before generative AI could create whole professions in one step.

141

Reposted by Lucie-Aimée Kaffee

Stella Biderman @stellaathena.bsky.social · 12/08/2025

Are you afraid of LLMs teaching people how to build bioweapons? Have you tried just... not teaching LLMs about bioweapons? @eleutherai.bsky.social‬ and the UK AISI joined forces to see what would happen, pretraining three 6.9B models for 500B tokens and producing 15 total models to study

412725

Lucie-Aimée Kaffee @frimelle.bsky.social · 11/08/2025

OpenAI just released GPT-5. When users share personal struggles, it sets fewer boundaries than o3. We tested both on INTIMA, our new benchmark for human-AI companionship behaviours. 🧵

151

Lucie-Aimée Kaffee @frimelle.bsky.social · 08/08/2025

Wikipedia has long been one of my favourite places online. As AI becomes part of knowledge creation, there's a lot we can learn from its editor communities. I spoke with Daniel Wu about AI content on Wikipedia; some thoughts made it into this piece: www.washingtonpost.com/technology/2...

washingtonpost.com

Volunteers fight to keep ‘AI slop’ off Wikipedia

Hundreds of Wikipedia articles may contain AI-generated errors. Editors are working around the clock to stamp them out.

073

Lucie-Aimée Kaffee @frimelle.bsky.social · 04/08/2025

New guide for open-source AI developers: Starting August 2, 2025, the EU AI Act imposes new rules on GPAI models, including open ones. What counts as GPAI? What’s exempt? What do you actually need to do? We wrote a guide (and built a tool) to help: huggingface.co/blog/yjernit...

huggingface.co

What Open-Source Developers Need to Know about the EU AI Act's Rules for GPAI models

A Blog post by Yacine Jernite on Hugging Face

030

Reposted by Lucie-Aimée Kaffee

Giada Pistilli @giadapistilli.com · 29/07/2025

From Replika to everyday chatbots, people form emotional bonds with AI. But what happens when an AI tells you "I understand how you feel" and you actually believe it? With @frimelle.bsky.social and @yjernite.bsky.social, we dug into something: how AI systems handle our emotional lives.

huggingface.co

AI Companionship: Why We Need to Evaluate How AI Systems Handle Emotional Bonds

A Blog post by Giada Pistilli on Hugging Face

143

Lucie-Aimée Kaffee @frimelle.bsky.social · 10/07/2025

Why was Grok taken down? No one knows for sure. But here’s the thing: You can flip a model’s entire vibe with just one line in the system prompt. Just ran this on @hf.co playground. Same question, two totally different answers 👇

100

Reposted by Lucie-Aimée Kaffee

Yacine Jernite @yjernite.bsky.social · 11/06/2025

Great blog post on *Digital Sovereignty and OS AI* led by the fantastic @frimelle.bsky.social! Digital sovereignty for AI needs to properly account for: 📚 data 🧑‍🔬 technology 💽 infrastructure ⚖️ regulation Open/transparent AI contributes to all, read for some concrete examples! hf.co/blog/frimell...

huggingface.co

Open Source AI: A Cornerstone of Digital Sovereignty

A Blog post by Lucie-Aimée Kaffee on Hugging Face

021

Lucie-Aimée Kaffee @frimelle.bsky.social · 11/06/2025

❗️New policy blogpost! The EU is speaking a lot about sovereignty. A cornerstone of digital sovereignty is and has to be open source. As AI becomes more central, the ability to govern, adapt, and understand these systems is no longer optional.

151

Reposted by Lucie-Aimée Kaffee

CPDPconferences @cpdpconferences.bsky.social · 05/05/2025

When Is a Summary of Data Used to Train GPAI Sufficiently Detailed? Organised by @ivir-uva.bsky.social with Kristina Irion (moderator), Zuzanna Warso, Yordanka Ivanova, @frimelle.bsky.social, Arriën Molema More information: cpdp.be/19190 #CPDP.ai #CPDP2025 #TheWorldisWatching #CPDP #GPAI

"When Is a Summary of Data Used to Train GPAI Sufficiently Detailed?"

It is an open secret that generative AI models have also been trained on any kind of datasets developers could get hold of and regardless of the legality of their practice. Concerns about the conformity with copyrights law and the General Data Protection Regulation abound. In relation to general-purpose AI (GPAI) models the European Union's Artificial Intelligence Act foresees the disclosure of public information about training data. Soon the providers of GPAI models will have to publish a sufficiently detailed summary of the data used to train their models. What this summary should look like is subject to ongoing debate as the EU’s AI Office is developing a template for the summary. In this panel we will take a closer look at the emerging contours of the template in light of the purpose of this summary.

Questions to be answered:
- What data information should the summary comprise?
- What does “sufficiently detailed” mean?
- How does trade secret protection factor in and does it affect the summary?
- And how does this summary enable parties with a legitimate interest to enforce their rights?

Organised by Institute for Information Law (IViR) with Kristina Irion (moderator), Zuzanna Warso, Yordanka Ivanova, Lucie-Aimée Kaffee, Arriën Molema

More information: https://cpdp.be/19190

022

Reposted by Lucie-Aimée Kaffee

CopeNLU @copenlu.bsky.social · 28/04/2025

In "Investigating Human Values in Online Communities", we perform a high-scale study of the unique values expressed by online communities with different perspectives arxiv.org/abs/2402.14177 #NAACL2025 #NLProc @nadavb.bsky.social @rnv.bsky.social @frimelle.bsky.social @iaugenstein.bsky.social

153

Reposted by Lucie-Aimée Kaffee

Hady Elsahar @hadyelsahar.bsky.social · 24/04/2025

Just met Chris Manning he said “we seem to be doing well” The younger me would have been proud hearing this. #iclr2025

001

Lucie-Aimée Kaffee @frimelle.bsky.social · 23/04/2025

I’m in Singapore this week for #ICLR2025 - co-organising the workshop on GenAI Watermaking sites.google.com/view/genai-w... and giving a keynote on policy & dataset transparency at the MLDPR workshop mldpr2025.com/schedule/

010

Lucie-Aimée Kaffee @frimelle.bsky.social · 13/03/2025

🚨 Our comments on the 3rd draft of the EU Code of Practice for GPAI developers are out! 🚨 Some promising updates—but also concerning choices that could harm transparency & open AI development. A quick breakdown of the blogpost with @yjernite.bsky.social 🧵👇

111

Reposted by Lucie-Aimée Kaffee

Yacine Jernite @yjernite.bsky.social · 12/03/2025

Amazing work from HF colleagues that's super relevant to AI policy - open development matches "frontier" performance 🤗 For policymakers, it means we need regulation that works for both open and closed model - not closed-first that disproportionately harms open! Good news is... 1/2

A graph comparing the performance of different AI systems on a benchmark derived from the Informatics Olympiads, showing an open Hugging Face model coming third ahead of DeepSeek-R1 and Claude-3.7 Sonnet

2138

Lucie-Aimée Kaffee @frimelle.bsky.social · 06/03/2025

Was sind eigentlich diese KI-Agenten, über die jetzt alle sprechen? 🤖 Die @dlfkultur-Folge zu KI Agenten erklärt es super! Ich durfte über die Rolle von Open Source sprechen; denn Transparenz & Anpassbarkeit sind entscheidend. 🎧 Hier anhören: www.deutschlandfunk.de/ki-agenten-k...

deutschlandfunk.de

Autonome KI - Wenn der Agent den Computer bedient

KI-Agenten übernehmen Recherchen, Buchungen, Terminabsprachen, Routineaufgaben. Können wir ihnen vertrauen? Und wieviel Macht sollten wir ihnen geben?

020

Lucie-Aimée Kaffee @frimelle.bsky.social · 20/02/2025

When I say my name, people start speaking French to me, although my French is basic. That also happens with AI systems. We wrote a whole paper on that, testing across models for presumed cultural identity based on names w/ Siddhesh Pawar @rnv.bsky.social @iaugenstein.bsky.social

2356

Lucie-Aimée Kaffee @frimelle.bsky.social · 13/02/2025

What a week in Paris with the AI Action Summit! From transparency and open source AI to sustainability and global alignment, the challenges ahead will require collaboration across policy makers, researchers and AI developers. On a more fun note, @hf.co had a huge party with amazing people 🤗

060

Lucie-Aimée Kaffee @frimelle.bsky.social · 12/02/2025

How should AI tools be designed to support rather than replace workers? At the Reshaping Work conference, I led a roundtable exploring AI’s impact on labor. We published a blogpost on our key takeaways on responsible AI and the future of work w/ Franco Bastida 🔗 www.rsm.nl/discovery/20... 🧵👇

rsm.nl

Start-Up Approaches to Responsible AI: Worker-Centric InnovationRotterdam school of Management, Erasmus University logoRotterdam school of Management, Erasmus University compact logo

Explore how start-ups are reshaping AI development through transparency, worker inclusivity, and ethical approaches that prioritise human augmentation over replacement.

153

Reposted by Lucie-Aimée Kaffee

Tzu-Sheng Kuo 郭子生 @tskuo.bsky.social · 06/02/2025

@wikipedia.org is vital for #NLP, yet little research contributes back to the Wikimedia community. To bridge this gap, we are organizing #WikiNLP to advance both. w/ Isaac Johnson (@wikimediafoundation.org) @akhilarora.bsky.social @frimelle.bsky.social @tiziano.bsky.social @indiiigo.bsky.social

0144

Lucie-Aimée Kaffee @frimelle.bsky.social · 04/02/2025

I was quoted in an article about the French Lucie AI. There were still a lot of problems with the model and how and when it was deployed. But seeing new smaller models being developed is an exciting direction for the next years of AI development to come! www.lapresse.ca/affaires/tec...

lapresse.ca

Radioscopie | Lucie, l’IA française qui ne passe pas le test

Elle parle comme Hitler et croit aux œufs de vache. Il n’en fallait pas plus pour que Lucie, l’IA générative française, soit ridiculisée par les internautes et retirée de la circulation.

000

Lucie-Aimée Kaffee @frimelle.bsky.social · 03/02/2025

Watching AI evolve has been a wild ride. From explaining why we'd bother generating a sentence with a neural net to clarifying that AI isn’t a magic, all-knowing box. Today, with the EU AI Act's first provisions in force, AI literacy isn’t optional. It’s foundational. 🧵

120

Reposted by Lucie-Aimée Kaffee

Sara Petti @sarapetti.bsky.social · 31/01/2025

It's so important not to discourage #data sharing but instead try to push for transparency when regulating #AI, says @frimelle.bsky.social on the last panel of the day on #OpenSourceAI. 👏👏👏

001

Reposted by Lucie-Aimée Kaffee

Eu/Acc @eu-acc.bsky.social · 29/01/2025

"High-quality efficient AI models are no longer the exclusive domain of tech giants with huge hardware resources. Europe could compete on “efficiency, specialisation and responsible AI development" with “small” AI models like DeepSeek" - @frimelle.bsky.social www.politico.eu/article/euro...

politico.eu

Europe’s AI hopes rebound after DeepSeek success

New Chinese model suggests winning the AI race could be done on the cheap.

053

Lucie-Aimée Kaffee @frimelle.bsky.social · 16/01/2025

🚨 Call for Papers: Workshop on Watermarking in Generative AI at @iclr-conf.bsky.social 🚨 Submit your work on watermarking, both technical work as well as discussions on advancements, security, evaluation & policy. 🤖🔏 Submission deadline: Feb 10, 2025 Details: sites.google.com/view/genai-w...

sites.google.com

ICLR 2025 Workshop on GenAI Watermarking

Dates

040

Lucie-Aimée Kaffee @frimelle.bsky.social · 14/01/2025

After a long holiday, finally back! 2025 goals I’m excited about: 🌱 Advocate for transparent & responsible AI development 💪 Balance work travel with home-cooked meals 🤝 Support diverse voices in AI legislation 📚 Research on AI policies and their technical implications

040

Reposted by Lucie-Aimée Kaffee

Molly White @molly.wiki · 08/01/2025

As I wrote last week: "As other information sources fall, Wikipedia’s stubborn independence becomes more vital than ever. The attacks from [the right] aren't just about an online encyclopedia — they're part of a broader assault on any information source that refuses to be controlled."

3531431335

Reposted by Lucie-Aimée Kaffee

Rob Pegoraro @robpegoraro.com · 31/12/2024

So glad you asked! The public-domain class of 2025 includes Hemingway’s A Farewell to Arms, Woolf’s A Room of One's Own, the 1.0 versions of Popeye and Tintin, Ravel’s “Boléro,” and the first recording of Gershwin’s “Rhapsody in Blue” - my favorite in this bunch.

pcmag.com

Hitchcock and Hemingway: All the Good Stuff Coming to the Public Domain in 2025

This year’s crop of creative works exiting copyright protection includes Gershwin’s 'An American In Paris,' Virginia Woolf’s A Room of One's Own, and the Marx Brothers’ feature-film debut.

8373141

Reposted by Lucie-Aimée Kaffee

Yacine Jernite @yjernite.bsky.social · 12/12/2024

...to immediate large-scale risks and to collaborative solutions supported by evidence helps - as long as developers disclose enough about their design choices. full blog here: hf.co/blog/yjernit... based on our submitted response with @frimelle.bsky.social and @brunatrevelin.bsky.social 🤗 2/2

hf.co

🇪🇺✍️ EU AI Act: Systemic Risks in the First CoP Draft Comments ✍️🇪🇺

A Blog post by Yacine Jernite on Hugging Face

042

Reposted by Lucie-Aimée Kaffee

Florent Daudens @fdaudens.bsky.social · 04/12/2024

📘 Just launched: Our Open Source Developer's Guide to the EU AI Act! Learn key compliance requirements & discover Hub tools to help you prepare. A must-read resource for the OS community. huggingface.co/blog/eu-ai-a... #EUAIAct #OpenSource

huggingface.co

Open Source Developers Guide to the EU AI Act

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

162

Lucie-Aimée Kaffee @frimelle.bsky.social · 03/12/2024

I appreciate so much being able to do policy work in the open for open-source AI, a topic that I think will affect all our lives. Finally added my responses to the EU AI stakeholder consultation and the US BIS response for @hf.co at huggingface.co/datasets/hug...

huggingface.co

huggingface/policy-docs · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

020

Lucie-Aimée Kaffee @frimelle.bsky.social · 02/12/2024

Excited to share our new article designed to help open source developers navigate the EU AI Act! You might be asking, "What does the EU AI Act mean for my projects?" We’ve just published a guide with @brunatrevelin.bsky.social @yjernite.bsky.social huggingface.co/blog/eu-ai-a...

huggingface.co

Open Source Developers Guide to the EU AI Act

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

152

Reposted by Lucie-Aimée Kaffee

Nathan Sarrazin @nsarrazin.com · 02/12/2024

Worked on a little UI for displaying the outputs of reasoning models on HuggingChat... Trying to find the right balance between letting users see the raw model output but also making the final answer easier to parse. QwQ can be *very* verbose so it seemed useful to give a summary at the end.

251

Lucie-Aimée Kaffee @frimelle.bsky.social · 28/11/2024

🚀 Excited to share that our provocation paper "Evaluations Using Wikipedia without Data Contamination: From Trusting Articles to Trusting Edit Processes" has been accepted at the NeurIPS workshop Evaluating Evaluations! 🌐📚 #NeurIPS2024 #evaleval #AIEvaluation

1204

Lucie-Aimée Kaffee @frimelle.bsky.social · 21/11/2024

'AI influencers' show how AI scales the exploitation of women’s images, monetising stolen content, erasing its origins. But let’s not forget that even if the images are not deep fakes, models are still trained on pictures of real women, without consent. www.wired.com/story/ai-pim...

wired.com

Inside the Booming ‘AI Pimping’ Industry

AI-generated influencers based on stolen images of real-life adult content creators are flooding social media.

220

Lucie-Aimée Kaffee @frimelle.bsky.social · 21/11/2024

Why do I prefer boring names for AI assistants? 🤔 Watch some snippets from Reshaping Work in Amsterdam last month, where I ran the roundtable discussion on "Supporting or replacing workers: How should AI tools be designed to benefit workers?" www.youtube.com/watch?v=h6Ol...

youtube.com

Unfiltered Insights from Reshaping Work 2024 – The Lighter Side of AI

YouTube video by Reshaping Work

020