Jinesis Lab (UToronto)

We are hosting a Dagstuhl seminar on Causality & LLMs this week (Apr 7–10). Bringing together world experts to explore: 1️⃣ Integrating LLMs 🤖 into causal workflows 2️⃣ Evaluating & improving LLMs’ causal reasoning 🧠 Co-organized w/ @amt_shrma @DominikJanzing @kunkzhang @ZhijingJin 📍Schloss Dagstuhl, Wadern, Germany 🔗 dagstuhl.de/26152 📖 cr-llm.github.io 📅 Apr 7–10 #CausalNLP #LLM #Dagstuhl @CausalNLP @MPI_IS @ELLISforEurope @UofTCompSci @VectorInst @TorontoSRI @CIFAR_News @JinesisLab @EuroSafeAI @ELLISInst_Tue Also joined with my student @rahulbshrestha to present our CauSciBench and Causal AI Scientist work :)!

📢We will present 5 papers to #ICLR2026, #CLeaR2026, and #ACL2026: - SocialHarmBench by @psyonp et al. - Causal LLMs on Instrumental Variable Method by @ivakshi_s et al. - LLM Data Contamination study by @TerryJCZhang et al. - Mech Interp for VLM by @francescortu et al. - DPO data selection method by Xuan & @rongwu_xu Thanks to all our collaborators and institutional support from @MPI_IS @ELLISforEurope @UofTCompSci @VectorInst @TorontoSRI @CIFAR_News @JinesisLab @EuroSafeAI @ELLISInst_Tue @ETH_en @ETH_AI_Center @michigan_AI @UMichiganAI @UMichCSE! Feel free to access the papers at arxiv.org/abs/2510.04891 arxiv.org/abs/2602.07943 arxiv.org/abs/2509.00072 arxiv.org/abs/2507.13868 arxiv.org/abs/2508.04149 🎉

What is the roadmap for NLP to actually help the world? 🌍 Thrilled to share our NLP for Social Good survey across nine domains, from healthcare and education to poverty, peacebuilding, and environmental protection. We analyze ACL Anthology trends and find that poverty, peacebuilding, and environmental protection remain underexplored. A call for cross-disciplinary partnerships and human-centered NLP, with 30+ authors! 📄 aclanthology.org/2026.eacl-long… #NLP4SG #EACL2026 #AI #ResponsibleAI

w/ @_AKassem, @bschoelkopf, @ZhijingJin With the support of @MPI_IS @ELLISforEurope @UofTCompSci @VectorInst @TorontoSRI @CIFAR_News @JinesisLab @ELLISInst_Tue

How robust are LLM routers, really? 🔀 We find that preference-based routers rely on category heuristics, not query complexity. They route ALL coding and math queries to the strongest LLM even when simpler models suffice, while sending jailbreaking attempts to weaker models, elevating safety risks! 🚨 We introduce the DSC benchmark: Diverse, Simple, and Categorized, evaluating routers across coding, math, translation, privacy, safety, and more. 📄 aclanthology.org/2026.eacl-long… #EACL2026 #AISafety #LLMs #NLP

with @YinyaHuang @MrinmayaSachan @ZhijingJin and the support of @MPI_IS @ELLISforEurope @JinesisLab @UofTCompSci @VectorInst @ETH_en @ETH_AI_Center

Standard text metrics miss correct LLM causal reasoning when the answer is valid but written differently. 🔍 Paul He presented DoVerifier: a symbolic verification framework using do-calculus and BFS with sound & complete guarantees! 🧩 📄 aclanthology.org/2026.eacl-long… #Causality #EACL2026 #NLP

w/ @keenansamway @NicoleMiuTakagi @Radamihalcea @bschoelkopf @KiddoThe2B @daniel_hers @ZhijingJin With the support of @MPI_IS @ELLISforEurope @UofTCompSci @VectorInst @TorontoSRI @CIFAR_News @JinesisLab @ELLISInst_Tue @michigan_AI @UMichCSE @uni_copenhagen @EuroSafeAI

When do LLMs decide that human rights principles should have "limitations"? 🚩 Our new paper explores how models endorse restrictions on fundamental rights. 📄 aclanthology.org/2026.findings-… #EACL2026 #HumanRights #AI #ResponsibleAI #NLP

w/ @JiaruiLiu Weihao Xuan @ZhijingJin @MonaDiab77 @MPI_IS @ELLISforEurope @UofTCompSci @VectorInst @TorontoSRI @CIFAR_News @JinesisLab @ELLISInst_Tue @LTIatCMU @RIKEN_AIP_EN

[Happening now] Catch our Oral! 🌮 📍 Salle Le Riad — Level 1 (Interpretability and Model Analysis) How do we stop Multimodal LLMs from seeing things that aren't there? We introduce TACO: a framework to mitigate object hallucinations through self-verification and atomic confidence calibration! 🖼️🚫 📄 aclanthology.org/2026.eacl-long… #EACL2026 #MLLMs #ResponsibleAI #ComputerVision

w/ @davidguzman1120 @strauss_irene @bschoelkopf @radamihalcea @ZhijingJin, supported by @MPI_IS @UofTCompSci @VectorInst @UMichCSE @michigan_AI @ETH_en @ETH_AI_Center @UZH_en !

Catch our Oral! 🏛️ 📍 Salle Walili — Level 0 (Measuring Social, Political, and Cultural Biases) Do LLMs lean democratic or authoritarian? We find models reject authoritarianism in English but score higher in Mandarin, even citing authoritarian role models! 🚩 📄 aclanthology.org/2026.eacl-long… #EACL2026 #ResponsibleAI #LLMs #AIBias

Kudos to the joint efforts by @psyonp @_yongjinny @Jiarui_Liu_ , supported by @UofTCompSci @MPI_IS @VectorInst @LTIatCMU !

How does competitive pressure change the way LLMs talk to each other? 🤖 Come check out our poster at #EACL2026! We introduce CORE: a new framework to measure multi-agent interaction quality when the stakes are high and game-theoretic pressures are on. Don't miss the discussion on AI coordination & safety! 📄 aclanthology.org/2026.eacl-long… #MultiAgent #Safety #NLP #GameTheory

Navigating mental health in the fast-paced world of AI research is a challenge we all face. 🧠 Join @strauss_irene and the @aclmentorship panel at #EACL2026 (Hybrid Rabat + Zoom) to discuss staying grounded. Submit/vote on questions here: app.sli.do/event/e4Em5p6f… #MentalHealth

AI is threatening our democratic society—by concentrating power, narrowing how we think, and flooding institutions faster than they can keep up. These risks emerge at the system level, and technical work alone won't fix them. 👉Check out our whitepaper with 25+ researchers: zhijing-jin.com/d/2026-ai-risk… 💡We introduce 7 threat models and ways forward. ✍️Led by @davidguzman1120 with @DaveRBanerjee, @blin_kevin, @PepijnCobben, @gcorsi_, @x_angelohuang, @ChanglingXavier, Suvajit Majumder, @psyonp, @SimkoSamuel, @strauss_irene, and @TerryJCZhang Advised by senior co-authors: @ashton1anderson, @Yoshua_Bengio, @MatthiasBethge, @RogerGrosse, Karoline Helbig, @david_lie, Richard Mallah, @radamihalcea, Susan Nesbitt, Susan Perry, @presnick, Stuart Russell, @mrinmayasachan, @bschoelkopf @audreyt and @ZhijingJin Thank you to all the institutional support from @JinesisLab @EuroSafeAI @MPI_IS @CIFAR_News @iapsAI @CARMA_411 @Cambridge_Uni @UofTCompSci @VectorInst @TorontoSRI @Mila_Quebec @LawZero_ @uni_tue @michigan_AI @UMichCSE @AUParis @UNESCO @UCBerkeley @ETH_en @ETH_AI_Center @ELLISInst_Tue @ELLISforEurope @EthicsInAI #CivicAI #AISafety #AIGovernance #Democracy #ResponsibleAI

🎉 Our lab has 7 papers at #EACL2026 in Rabat this week 🇲🇦 Topics span democracy defense, multi agent safety, causal reasoning, hallucinations, and NLP for social good. Grateful to everyone who contributed to this work 🙌 🙌 Come find us! #NLProc #LLMs #ResponsibleAI

Difficult times—but we keep pushing forward. ✅ Our Trustworthy AI for Good Workshop→accepted at @icmlconf Seoul (18% acceptance) ✅ @NLP4PosImpact Workshop→coming to @emnlpmeeting in Budapest 🇭🇺 AI research can be a force for good—and we’re committed to contribute. More soon.