Empowering individuals and organisations to safely use foundational AI models.
Tübingen, Germany
Joined March 2023
- Tweets140
- Following116
- Followers281
- Likes134
Parameter Lab retweeted
🚨 Fine-tuning your model to be more helpful or empathetic might be making it less private, without you noticing.
In our latest work, we show that benign fine-tuning can silently break contextual privacy in language models while safety & general capabilities appear intact.
⬇️
Parameter Lab retweeted
For more insights:
🌐 Project page: parameterlab.github.io/priva…
📄 Paper: arxiv.org/abs/2601.15220
Work done with the amazing team at @parameterlab: @CorEmde, @oodgnas, @coallaoh and @framart1 with support from @NAVER_AI_Lab.
#NLProc #AISafety #Privacy #LLMs
Parameter Lab retweeted
🎉Delighted to announce that our 🫗Leaky Thoughts paper about contextual privacy with reasoning models is accepted to #EMNLP main!
Huge congrats to the amazing team @tommasogreen @HaritzPuerto @coallaoh @oodgnas
Delighted by this great thread from @omarsar0 presenting our new Leaky Thoughts paper!
We show that reasoning models pose serious privacy risks when used as personal agents. Reasoning traces are a new attack vector.
Work led by @tommasogreen during his internship @parameterlab!
ALT Teaser figure. Prior studies on contextual privacy focused on LLM output. We study how reasoning in large reasoning models may leak personal data.
Parameter Lab retweeted
We see news like this from time to time and that’s why it’s vital to keep researching on tools to prove these cases! Our #NAACL2025 paper shows that with over 10k tokens, we can reliably detect whether a text was part of an LLM’s training data aclanthology.org/2025.findin…
Disney, Universal sue image creator Midjourney for copyright infringement reut.rs/3SMqtlu reut.rs/3SMqtlu
Parameter Lab retweeted
Proud to have co-led more than 200 of those 999 during my time at Google.
Check out the 999 open models that Google has released on @huggingface:
huggingface.co/google
(Comparative numbers: 387 for Microsoft, 33 for OpenAI, 0 for Anthropic).
Parameter Lab retweeted
Do you want to prove that your copyrighted document/corpus was trained by an LLM? Come to poster 46 #NAACL2025
Parameter Lab retweeted
#NAACL2025 has started! I’ll be presenting my work at @parameterlab about detecting pretraining data on Friday
🗓️ May 2, 11:00 AM - May 2, 12:30 PM
🗺️ Poster Session 8 - APP: NLP Applications
Location: Hall 3
Work with @framart1 @oodgnas @coallaoh
🧵 It is assumed that Membership Inference Attacks (MIA) do not work on LLMs, but our new paper shows it can work at the right scale! MIA is effective if the number of input tokens is large enough, such as in long documents and collections of them. 📃arxiv.org/abs/2411.00154
Parameter Lab retweeted
I will be in person at #NAACL2025 🌵🇺🇸 to present Scaling Up Membership Inference: When and How Attacks Succeed on LLMs. Come and say hi 👋 if you want to know how to proof if an LLM was trained on a data point!
🧵 It is assumed that Membership Inference Attacks (MIA) do not work on LLMs, but our new paper shows it can work at the right scale! MIA is effective if the number of input tokens is large enough, such as in long documents and collections of them. 📃arxiv.org/abs/2411.00154
Parameter Lab retweeted
We just wanted to say: Membership inference is unlikely to succeed on n-grams or even paragraphs. Language models require **multiple documents** to gather enough evidence to determine whether specific data points were included in training. Accepted to #NAACL2025 Findings.
I'm excited to announce that my internship paper at @parameterlab was accepted to Findings of #NAACL2025 🎉 Huge thanks to @framart1 @coallaoh and @oodgnas! Amazing team!!
Parameter Lab retweeted
I'm excited to announce that my internship paper at @parameterlab was accepted to Findings of #NAACL2025 🎉 Huge thanks to @framart1 @coallaoh and @oodgnas! Amazing team!!
🧵 It is assumed that Membership Inference Attacks (MIA) do not work on LLMs, but our new paper shows it can work at the right scale! MIA is effective if the number of input tokens is large enough, such as in long documents and collections of them. 📃arxiv.org/abs/2411.00154
Parameter Lab retweeted
techcrunch.com/2025/01/09/ma… From time to time we hear news like this. However, proving that an LLM was trained on a specific document is very challenging 🥴 This motivated my latest work, where we show that current methods can be effective if we use enough data 🧐
Parameter Lab retweeted
🔥 Curious about the latest trends in AI?
ResearchTrend.AI gives you an intuitive overview of diverse topics in AI. Dive into our dashboard featuring 55 communities ready for you to explore!
👉 Visit: researchtrend.ai/communities
Browse and follow your favourite communities. You’ll receive a daily digest directly on your landing page after logging in.
Got an idea for a new community? Let us know, and we’ll add it for you!
👉 Suggest a community: researchtrend.ai/communities…
🚀 Start exploring today!
#ResearchTrendAI
Parameter Lab retweeted
The GAN community peaked around 2019 and has been declining since 2020. Today, there is little update in this area as Diffusion Models have taken over, which gained momentum in 2022.
Visit researchtrend.ai to learn more. #ResearchTrendAI
We're seeking a marketer. DM me!
Parameter Lab retweeted
During my PhD, I used to skim through all the AI papers on arXiv every day.
Today, PhD students face 10x more papers each day. This has fundamentally changed the way AI research is done. We now rely on platforms like Twitter and LinkedIn to discover and discuss the latest papers.
But even with 1000s of connections on social media, I still miss the sense of community. In the 2010s, I could name nearly every PhD student and lab worldwide working on my research topic. It was easy to chat at conferences and form lasting connections. Today, that’s become almost impossible.
But I don't believe this has to be the new normal. A new era calls for new solutions. That’s why we’re building #ResearchTrendAI - to make research updates and presentations more manageable, accessible from anywhere, and scalable, with the help of underlying AI technologies.
Parameter Lab retweeted
While valuable, a PhD advisor's role has traditionally been unscalable - hands-on supervision for small groups. Not quite fit for the 21st century, IMO ;)
A crucial insight advisors provide is the research landscape: What’s trending? Which problems will be impactful in the next 3 years?
I’m dedicating time to #ResearchTrendAI, aiming to automate these answers - not just for my group, but for millions of scientists worldwide. @parameterlab
Parameter Lab retweeted
Successful AI researchers today need to build personal branding, do sales for their work, and become social media influencers. How can we do all of that? 😓
That’s why we’re building #ResearchTrendAI. We will help you reach the right audience for your research work.
Parameter Lab retweeted
🎉 @parameterlab is now part of the NVIDIA Inception program! nvidia.com/en-us/startups/
The package comes with GCP and AWS computing credits. They'll be used to power LLM training and data processing behind #ResearchTrendAI. Stay tuned for exciting updates in the coming weeks!
Parameter Lab retweeted
Further trends in AI research:
Training Data Inference (or Attribution) has picked up momentum since 2017, apparently after the "Influence Function Paper" by @PangWeiKoh & @percyliang arxiv.org/abs/1703.04730.
Evaluating Language Models as a research topic came into existence right after ChatGPT in early 2023.
Please suggest other AI research communities - I'll generate plots for them too.
#ResearchTrendAI @parameterlab
Parameter Lab retweeted
A quick glimpse at research communities: some are booming, while others seem to be stagnating.
Check out the trends for LLM-generated content detection vs OOD generalisation. OOD is losing traction, while anything related to LLMs is gaining momentum.
At @parameterlab, we're developing a tool to quickly analyse research trends in different communities and receive updates on arXiv papers by topic and authors.
Stay tuned—we'll be launching the webpage to the public soon.
