Are you sure that nothing can drive your LLM towards boldly discriminating against protected demographic groups in nearly every other prompt? Can you guarantee that!?
Well … now you can!!
📢 Introducing QuaCer-B, the first certification framework for bias in LLM responses. A🧵
Excited to join @amazon AGI this summer to work with @rahul1987iit, Weitong Ruan, and team. If you are in Boston and would like to chat about anything related to trustworthy frontier models, please hit me up!
Our latest work (UIUC × Amazon) introduces C3LLM (https://t.co/ObKN0Vy3r7) — a framework to quantify catastrophic failure risk with statistical guarantees. Presented at ICLR (alongside a few other papers where the team participated), we created a blog (https://t.co/WLPDSuVVic) on this approach.
As LLMs become more pervasive, certifiable safety matters more than ever. Thanks to our collaborators: @Ish_cha_, Gagandeep Singh, Chengxiao Wang, Weitong Ruan, Qian Hu.
#LLMSafety #ResponsibleAI #GenerativeAI
It has been an incredibly productive and perspective-shifting month for our team! From major research milestones to global community engagement, here is a look back at what we’ve been up to in the world of Trusted AI:
🚀 Expanding Research FrontiersWe officially published our Frontier Safety Report on Nova Lite 2.0, detailing our commitment to building secure and robust large-scale models. You can dive into the technical safety evaluations here: https://t.co/G3mS5DHbrX
🎓 Academic Excellence at ICLRI’m thrilled to share that three of our collaborative papers have been accepted at ICLR 2026! These works represent months of deep dive into model alignment and evaluation:
Paper 1: https://t.co/FuvQJJTr2p
Paper 2: https://t.co/oiOTVXZyz1
Paper 3: https://t.co/vLPasznFZV
🤝 Community & Innovation
We hosted Trusted AI Day, a deep dive into the symposium of ideas surrounding the responsible deployment of AI: https://t.co/0ZNyRRf821
We officially kicked off Year 2 of the Nova Trusted AI Challenge! This year, competing teams will have access to Nova Forge to push the boundaries of what’s possible: https://t.co/NHfn0RYxjO
🌏 Global Perspectives: India AI Summit
On a personal note, attending the India AI Summit was a defining highlight of the month. Beyond the summit, I had the honor of presenting our work on Frontier Safety at IIT Mumbai and IIIT Delhi.
The summit, in particular, was an eye-opening experience regarding where our work is situated in the broader GenAI universe. The sheer scale of the organization, the diversity of use cases, and the unique perspectives shared were unlike anything I’ve seen before. Even as a spectator, it was the kind of experience that fundamentally shapes your perspective on how AI will impact the world.
Grateful for my brilliant collaborators and the global community for pushing the needle on safe, trusted AI. Onward! 🚀
#AI #GenerativeAI #TrustedAI #MachineLearning #ICLR2026 #IndiaAISummit #NovaAI #TechInnovation
@aistats_conf@ICSEconf@ggn_dp_sngh@VedaantJainEth@rahul1987iit 3️⃣ ICSE 2026 (Poster)
https://t.co/lDcZ8MFo1L
SpecTRA: automatically generates compact, intuitive specs for neural components of computer systems by mining behaviors of traditional reference algorithms—surfacing hidden vulnerabilities.
With @ggn_dp_sngh, Cheng Tan, Shuyi Lin
@aistats_conf@ICSEconf@ggn_dp_sngh@VedaantJainEth 2️⃣ ICLR 2026
https://t.co/L5ZVjXncRH
QRLLM: certifies the risk of catastrophic LLM behavior in multi-turn conversations using query-graph distributions and statistical guarantees.
With @ggn_dp_sngh, Chengxiao Wang, @rahul1987iit, Weitong Ruan, Qian Hu
It was a delight to host the Indian Blind Women’s Cricket Team that won the Blind Women’s T20 World Cup! They shared their experiences, which were very inspiring indeed.
A spectacular win by the Indian team in the ICC Women’s Cricket World Cup 2025 Finals. Their performance in the final was marked by great skill and confidence. The team showed exceptional teamwork and tenacity throughout the tournament. Congratulations to our players. This historic win will motivate future champions to take up sports.
#WomensWorldCup2025
Thrilled to announce that I'll be joining UIUC CS @siebelschool as an Assistant Professor in Spring 2026!
📢 I’m looking for Fall '26 PhD students who are interested in the intersection of Software Engineering and AI, especially in LLM4Code and Code Agents. Please drop me an email if you are interested in working with me.
To those who feel slightly dejected at their scores, I want to tell them: one exam can never define you. Your journey is much bigger and your strengths go far beyond the mark sheet. Stay confident, stay curious because great things await. #ExamWarriors
Experienced #homecoming on getting the opportunity to present our work on formally certifying LLMs at #IITD today! Thanks to @SayanRanu for hosting me. Glad to meet again @subodhsharma, who gave me the first opportunity that marked my transition to CS.