Iacopo Masi

We finally did it with 5/5 papers accepted at #ICLR2026 🦾🇧🇷 Thanks to my @OmnAI_Lab group and my collaborators for this amazing accomplishment! See you in Rio!

@karpathy have you seen Flywheel? this is basically what we're building. x.com/i/status/20321…

Drifting models as score-based models!

Why can't we solve adversarial examples? After a decade of work, neural nets still get fooled by imperceptible noise. We think we finally know the geometric reason why — and it connects to AI alignment. 🧵

hypemaxxing

#ECCV2026 🇸🇪 received 10,665 submissions going to review. That is an ECCV record. +24% vs #ECCV2024 (8,585) #ECCV2022: 6,773

we're officially live with the open beta of Flywheel. Flywheel is Paradigma's vision on what autonomous research will run on. We've built all of this pre-funding, so expect the pace to accelerate significantly. Super eager to hear your feedback and your ideas. I'll be live tweeting some of my experiments with Flywheel in a few minutes. We're just getting started.

@mathtician @OmnAI_Lab Yes the gist is that there are hidden degrees of freedom at each autoregressive step that the model can choose but these quantities in principle should be equal across two time steps to model properly the energy of a sentence E(x1,..,xN).

@OmnAI_Lab Really nice! Idk why we expect the spilled energy at each token to be 0, though, since the NLL of a sequence is the sum of spilled energies plus some O(1) boundary terms. Maybe all spilled energies should be the avg entropy/token? The histograms in Fig. 3 have some positive bias.

1/ Large Language Models leak energy when they hallucinate. We built a training-free method to catch the spill and keep them *grounded*. Our #ICLR2026 paper introduces Spilled Energy for SOTA zero-shot detection. TLDR: Hallucinations violate the probability chain rule.

#IEEE #FG2026 registration is now OPEN! Register now at fg2026.ieee-biometrics.org/registration/

@OmnAI_Lab What we also find very interesting is: - instruct variants show a higher spillage to hallucination correlation; post-training artifacts? - the math behind does not apply to just language modeling, any sequence-to-sequence task abides by the same rules!

Spilled Energy in Large Language Models Adrian Robert Minut, Hazem Dewidar, Iacopo Masi arxiv.org/abs/2602.18671 [𝚌𝚜.𝙰𝙸 𝚌𝚜.𝙲𝙻]

Terence Tao says AI is shifting mathematics from isolated case studies to large-scale projects Instead of spending months on a single problem, researchers can now analyze thousands at once As "citizen mathematics" grows and AI starts delivering real value, the field is opening up far beyond just PhDs

The Call for Papers for Machine Unlearning for Vision @CVPR is officially open! We’re looking for work on making vision models forget — safely, efficiently, and at scale. Consider submitting! #CVPR #Unlearning #xAI #CV

CVPR 2026 decisions are out 📢 Congrats to all accepted papers 🎉 If your work didn’t make the main conference but you believe in the idea, consider submitting to MUV 🧠🔐 Accepted? Share it via our non-proceedings track. …chine-unlearning-for-vision.github.io #CVPR2026 @CVPR

We welcome submissions on topics including: • Machine unlearning & data removal • Concept erasure & model editing • Safety, robustness & reliability in vision • Debiasing and governance of foundation models • Evaluation protocols for unlearning 📝 Submit by March 15th!

Don’t miss MUV @CVPR 2026! 🚨 If your work touches machine unlearning or safe adaptation in vision, this is your venue. 📝 Submit by March 15. #CVPR2026 #MUV #Workshop

A really dangerous situation. Too many submissions. Too many generated papers. Little responsibility. 1. In 2026, more than 24,000 submissions were made to the International Conference on Machine Learning (ICML). It’s TWO times more than in 2025. To fight it, the organizers now require researchers to pay $100 for every subsequent paper. 2. LLM adoption has increased researcher productivity by 90% (there’s a recent paper in Science). 3. The number of papers is becoming far too high. Submissions to arXiv have risen by 50% since 2022. 4. There are simply not enough reviewers. Plus, many scientists no longer want to invest precious time in it for free. 5. We can’t easily identify AI-made papers from the genuine ones. __ Important words from Paul Ginsparg, a co-founder of arXiv: “AI slop frequently can’t be discriminated just by looking at abstract, or even by just skimming full text. This makes it an “existential threat” to the system.” Basically, we’re getting closer to the tipping point. 📍 Many professors blame the AI. But the problem is likely elsewhere: 1. Without a sufficient number of papers, many PIs can’t get funded. They have to prove their credibility to reviewers. Their proposals have to rely on prior publications. In many countries, there are some informal (or even formal) expectations for how many papers a group with a certain size has to publish to survive (funding-wise). 2. Our students / postdocs need papers if they want to be hired in faculty roles. Yes, some departments hire people with few publications. But the majority still want to ensure their faculty can get funded. If funding is partly a function of papers, this is used in decision-making. 3. The number of papers is important if you want to get high-level awards. Many of them are not given because you published one paper (even if it’s great). They are given because you made a meaningful CONTRIBUTION to the field. How do you make it? Publish more papers. 4. Tenure promotions in many places take the number of your papers into account (often indirectly). Your tenure may get delayed if you don’t publish enough. Not everywhere, but for many mid- to low-ranked universities this story is more or less the same. + There are many more to mention. 📍My opinion: Much of this is rooted in how funding is distributed. There is a strong correlation between the requirements at a university and the funding acquisition criteria. If funding were based ONLY on the quality of published papers, universities would hire people for the quality of their science. If funding agencies strongly discouraged publishing too many papers, universities wouldn’t expect numbers from faculty during promotions. And some supervisors wouldn’t pressure students and postdocs to publish unfinished studies and low-quality data. Yes, we need good detectors of fake papers. But we also need the right policies and better funding allocation criteria.

Can AI forget? 🧠❌ Join MUV at @CVPR 26 in Denver! 🏔️ Speakers from @GoogleDeepMind, @MIT_CSAIL & more. 📝 Submit by March 15! Organizers: @SapienzaRoma, @MIT, @TU_Muenchen, @_italai and MPI. Details: …chine-unlearning-for-vision.github.io #CVPR2026 #AI #ComputerVision

Assuming AGI becomes widely available, what is the future of science? We are launching the P-AGI workshop at @iclr_conf to define the post-AGI research agenda @DonatoCrisosto1 @teelinsan @valentina__py @pratyusha_PS @ZorahLaehner @EmanueleRodola

👋🏼Hi Fellas, if you work on machine unlearning for vision and related topics, please consider taking a look at our brand new @CVPR 26 workshop 🦾 …chine-unlearning-for-vision.github.io with @AlessioSampier1 @bardh95 @GalassoFab10 @materzynska and Bernt Schiele 🙏🏼