Wiredframe Radar logo

Category intelligence

Research Briefing — January 8, 2026

395 current items analyzed and ranked.

Executive synthesis

Research Summary

The OpenAI GPT-5 System Card dominates today's releases, detailing the unified architecture with dynamic routing between fast and deep reasoning modes. A remarkable autoformalization result shows 130k lines of formal topology generated in two weeks for ~$100, suggesting accessible mathematical formalization at scale.

Safety and alignment research features prominently:

Theoretical and interpretability advances challenge key assumptions:

Key Themes

Language Models · 56AI Safety & Alignment · 44AI Agents · 11Reasoning & Chain-of-Thought · 12Reinforcement Learning for LLMs · 14AI Safety & Security · 10Benchmarks & Evaluation · 22Scientific AI · 6Multimodal Models · 11Mechanistic Interpretability · 6

Primary evidence

Top Ranked Signals

Research arXiv (Computation and Language) Jan 8

OpenAI GPT-5 System Card

By Aaditya Singh, Adam Fry, Adam Perelman, Adam Tart, Adi Ganesh, Ahmed El-Kishky, Aidan McLaughlin, Aiden Low, AJ Ostrow, Akhila Ananthram, Akshay Nathan, Alan Luo, Alec Helyar, Aleksander Madry, Aleksandr Efremov, Aleksandra Spyra, Alex Baker-Whitcomb, Alex Beutel, Alex Karpenko, Alex Makelov, Alex Neitz, Alex Wei, Alexandra Barr, Alexandre Kirchmeyer, Alexey Ivanov, Alexi Christakis, Alistair Gillespie, Allison Tam, Ally Bennett, Alvin Wan, Alyssa Huang, Amy McDonald Sandjideh, Amy Yang, Ananya Kumar, Andre Saraiva, Andrea Vallone, Andrei Gheorghe, Andres Garcia Garcia, Andrew Braunstein, Andrew Liu, Andrew Schmidt, Andrey Mereskin, Andrey Mishchenko, Andy Applebaum, Andy Rogerson, Ann Rajan, Annie Wei, Anoop Kotha, Anubha Srivastava, Anushree Agrawal, Arun Vijayvergiya, Ashley Tyra, Ashvin Nair, Avi Nayak, Ben Eggers, Bessie Ji, Beth Hoover, Bill Chen, Blair Chen, Boaz Barak, Borys Minaiev, Botao Hao, Bowen Baker, Brad Lightcap, Brandon McKinzie, Brandon Wang, Brendan Quinn, Brian Fioca, Brian Hsu, Brian Yang, Brian Yu, Brian Zhang, Brittany Brenner, Callie Riggins Zetino, Cameron Raymond, Camillo Lugaresi, Carolina Paz, Cary Hudson, Cedric Whitney, Chak Li, Charles Chen, Charlotte Cole, Chelsea Voss, Chen Ding, Chen Shen, Chengdu Huang, Chris Colby, Chris Hallacy, Chris Koch, Chris Lu, Christina Kaplan, Christina Kim, CJ Minott-Henriques, Cliff Frey, Cody Yu, Coley Czarnecki, Colin Reid, Colin Wei, Cory Decareaux, Cristina Scheau, Cyril Zhang, Cyrus Forbes, Da Tang, Dakota Goldberg, Dan Roberts, Dana Palmie, Daniel Kappler, Daniel Levine, Daniel Wright, Dave Leo, David Lin, David Robinson, Declan Grabb, Derek Chen, Derek Lim, Derek Salama, Dibya Bhattacharjee, Dimitris Tsipras, Dinghua Li, Dingli Yu, DJ Strouse, Drew Williams, Dylan Hunn, Ed Bayes, Edwin Arbus, Ekin Akyurek, Elaine Ya Le, Elana Widmann, Eli Yani, Elizabeth Proehl, Enis Sert, Enoch Cheung, Eri Schwartz, Eric Han, Eric Jiang, Eric Mitchell, Eric Sigler, Eric Wallace, Erik Ritter, Erin Kavanaugh, Evan Mays, Evgenii Nikishin, Fangyuan Li, Felipe Petroski Such, Filipe de Avila Belbute Peres, Filippo Raso, Florent Bekerman, Foivos Tsimpourlas, Fotis Chantzis, Francis Song, Francis Zhang, Gaby Raila, Garrett McGrath, Gary Briggs, Gary Yang, Giambattista Parascandolo, Gildas Chabot, Grace Kim, Grace Zhao, Gregory Valiant, Guillaume Leclerc, Hadi Salman, Hanson Wang, Hao Sheng, Haoming Jiang, Haoyu Wang, Haozhun Jin, Harshit Sikchi, Heather Schmidt, Henry Aspegren, Honglin Chen, Huida Qiu, Hunter Lightman, Ian Covert, Ian Kivlichan, Ian Silber, Ian Sohl, Ibrahim Hammoud, Ignasi Clavera, Ikai Lan, Ilge Akkaya, Ilya Kostrikov, Irina Kofman, Isak Etinger, Ishaan Singal, Jackie Hehir, Jacob Huh, Jacqueline Pan, Jake Wilczynski, Jakub Pachocki, James Lee, James Quinn, Jamie Kiros, Janvi Kalra, Jasmyn Samaroo, Jason Wang, Jason Wolfe, Jay Chen, Jay Wang, Jean Harb, Jeffrey Han, Jeffrey Wang, Jennifer Zhao, Jeremy Chen, Jerene Yang, Jerry Tworek, Jesse Chand, Jessica Landon, Jessica Liang, Ji Lin, Jiancheng Liu, Jianfeng Wang, Jie Tang, Jihan Yin, Joanne Jang, Joel Morris, Joey Flynn, Johannes Ferstad, Johannes Heidecke, John Fishbein, John Hallman, Jonah Grant, Jonathan Chien, Jonathan Gordon, Jongsoo Park, Jordan Liss, Jos Kraaijeveld, Joseph Guay, Joseph Mo, Josh Lawson, Josh McGrath, Joshua Vendrow, Joy Jiao, Julian Lee, Julie Steele, Julie Wang, Junhua Mao, Kai Chen, Kai Hayashi, Kai Xiao, Kamyar Salahi, Kan Wu, Karan Sekhri, Karan Sharma, Karan Singhal, Karen Li, Kenny Nguyen, Keren Gu-Lemberg, Kevin King, Kevin Liu, Kevin Stone, Kevin Yu, Kristen Ying, Kristian Georgiev, Kristie Lim, Kushal Tirumala, Kyle Miller, Lama Ahmad, Larry Lv, Laura Clare, Laurance Fauconnet, Lauren Itow, Lauren Yang, Laurentia Romaniuk, Leah Anise, Lee Byron, Leher Pathak, Leon Maksin, Leyan Lo, Leyton Ho, Li Jing, Liang Wu, Liang Xiong, Lien Mamitsuka, Lin Yang, Lindsay McCallum, Lindsey Held, Liz Bourgeois, Logan Engstrom, Lorenz Kuhn, Louis Feuvrier, Lu Zhang, Lucas Switzer, Lukas Kondraciuk, Lukasz Kaiser, Manas Joglekar, Mandeep Singh, Mandip Shah, Manuka Stratta, Marcus Williams, Mark Chen, Mark Sun, Marselus Cayton, Martin Li, Marvin Zhang, Marwan Aljubeh, Matt Nichols, Matthew Haines, Max Schwarzer, Mayank Gupta, Meghan Shah, Melody Huang, Meng Dong, Mengqing Wang, Mia Glaese, Micah Carroll, Michael Lampe, Michael Malek, Michael Sharman, Michael Zhang, Michele Wang, Michelle Pokrass, Mihai Florian, Mikhail Pavlov, Miles Wang, Ming Chen, Mingxuan Wang, Minnia Feng, Mo Bavarian, Molly Lin, Moose Abdool, Mostafa Rohaninejad, Nacho Soto, Natalie Staudacher, Natan LaFontaine, Nathan Marwell, Nelson Liu, Nick Preston, Nick Turley, Nicklas Ansman, Nicole Blades, Nikil Pancha, Nikita Mikhaylin, Niko Felix, Nikunj Handa, Nishant Rai, Nitish Keskar, Noam Brown, Ofir Nachum, Oleg Boiko, Oleg Murk, Olivia Watkins, Oona Gleeson, Pamela Mishkin, Patryk Lesiewicz, Paul Baltescu, Pavel Belov, Peter Zhokhov, Philip Pronin, Phillip Guo, Phoebe Thacker, Qi Liu, Qiming Yuan, Qinghua Liu, Rachel Dias, Rachel Puckett, Rahul Arora, Ravi Teja Mullapudi, Raz Gaon, Reah Miyara, Rennie Song, Rishabh Aggarwal, RJ Marsan, Robel Yemiru, Robert Xiong, Rohan Kshirsagar, Rohan Nuttall, Roman Tsiupa, Ronen Eldan, Rose Wang, Roshan James, Roy Ziv, Rui Shu, Ruslan Nigmatullin, Saachi Jain, Saam Talaie, Sam Altman, Sam Arnesen, Sam Toizer, Sam Toyer, Samuel Miserendino, Sandhini Agarwal, Sarah Yoo, Savannah Heon, Scott Ethersmith, Sean Grove, Sean Taylor, Sebastien Bubeck, Sever Banesiu, Shaokyi Amdo, Shengjia Zhao, Sherwin Wu, Shibani Santurkar, Shiyu Zhao, Shraman Ray Chaudhuri, Shreyas Krishnaswamy, Shuaiqi (Tony) Xia, Shuyang Cheng, Shyamal Anadkat, Sim\'on Posada Fishman, Simon Tobin, Siyuan Fu, Somay Jain, Song Mei, Sonya Egoian, Spencer Kim, Spug Golden, SQ Mah, Steph Lin, Stephen Imm, Steve Sharpe, Steve Yadlowsky, Sulman Choudhry, Sungwon Eum, Suvansh Sanjeev, Tabarak Khan, Tal Stramer, Tao Wang, Tao Xin, Tarun Gogineni, Taya Christianson, Ted Sanders, Tejal Patwardhan, Thomas Degry, Thomas Shadwell, Tianfu Fu, Tianshi Gao, Timur Garipov, Tina Sriskandarajah, Toki Sherbakov, Tomer Kaftan, Tomo Hiratsuka, Tongzhou Wang, Tony Song, Tony Zhao, Troy Peterson, Val Kharitonov, Victoria Chernova, Vineet Kosaraju, Vishal Kuo, Vitchyr Pong, Vivek Verma, Vlad Petrov, Wanning Jiang, Weixing Zhang, Wenda Zhou, Wenlei Xie, Wenting Zhan, Wes McCabe, Will DePue, Will Ellsworth, Wulfie Bain, Wyatt Thompson, Xiangning Chen, Xiangyu Qi, Xin Xiang, Xinwei Shi, Yann Dubois, Yaodong Yu, Yara Khakbaz, Yifan Wu, Yilei Qian, Yin Tat Lee, Yinbo Chen, Yizhen Zhang, Yizhong Xiong, Yonglong Tian, Young Cha, Yu Bai, Yu Yang, Yuan Yuan, Yuanzhi Li, Yufeng Zhang, Yuguang Yang, Yujia Jin, Yun Jiang, Yunyun Wang, Yushi Wang, Yutian Liu, Zach Stubenvoll, Zehao Dou, Zheng Wu, Zhigang Wang

98 score
AI Analysis
Official system card for OpenAI GPT-5 launch (August 2025), describing a unified system with fast and deep reasoning models, real-time router for complexity-based model selection, and continuous training on user feedback signals.
This is the system card published alongside the OpenAI GPT-5 launch, August 2025. GPT-5 is a unified system with a smart and fast model that answers most questions, a deeper reasoning model for harder problems, and a real-time router that quickly decides which model to use based on conversation type, complexity, tool needs, and explicit intent (for example, if you say 'think hard about this' in the prompt). The router is continuously trained on real signals, including when users switch models,
Language ModelsAI SafetyOpenAIFoundation Models
Research arXiv (cs.LO) Jan 8

130k Lines of Formal Topology in Two Weeks: Simple and Cheap Autoformalization for Everyone?

By Josef Urban

88 score
AI Analysis
Reports autoformalization of 130k lines of topology from Munkres textbook in two weeks for ~$100 LLM cost, including proofs of Urysohn's lemma and metrization theorem, using LLM-proof checker feedback loop.
This is a brief description of a project that has already autoformalized a large portion of the general topology from the Munkres textbook (which has in total 241 pages in 7 chapters and 39 sections). The project has been running since November 21, 2025 and has as of January 4, 2026, produced 160k lines of formalized topology. Most of it (about 130k lines) have been done in two weeks,from December 22 to January 4, for an LLM subscription cost of about \$100. This includes a 3k-line proof of Urys
Formal MethodsLanguage ModelsMathematicsAutoformalization
Research arXiv (Computation and Language) Jan 8

Jailbreak-Zero: A Path to Pareto Optimal Red Teaming for Large Language Models

By Kai Hu, Abhinav Aggarwal, Mehran Khodabandeh, David Zhang, Eric Hsin, Li Chen, Ankit Jain, Matt Fredrikson, Akash Bharadwaj

82 score
AI Analysis
Introduces Jailbreak-Zero, a red teaming methodology shifting from example-based to policy-based LLM safety evaluation. Uses attack LLM fine-tuned with preference data to achieve Pareto optimality across coverage, diversity, and fidelity, showing high success rates against GPT-4o and Claude 3.5.
This paper introduces Jailbreak-Zero, a novel red teaming methodology that shifts the paradigm of Large Language Model (LLM) safety evaluation from a constrained example-based approach to a more expansive and effective policy-based framework. By leveraging an attack LLM to generate a high volume of diverse adversarial prompts and then fine-tuning this attack model with a preference dataset, Jailbreak-Zero achieves Pareto optimality across the crucial objectives of policy coverage, attack strateg
AI SafetyRed TeamingLanguage ModelsAlignment
Research arXiv (Artificial Intelligence) Jan 8

Mastering the Game of Go with Self-play Experience Replay

By Jingbin Liu and Xuechun Wang

82 score
AI Analysis
Introduces QZero, a model-free RL algorithm for Go that learns Nash equilibrium policy through self-play and experience replay without MCTS, achieving AlphaGo-level performance with modest compute (7 GPUs, 5 months).
The game of Go has long served as a benchmark for artificial intelligence, demanding sophisticated strategic reasoning and long-term planning. Previous approaches such as AlphaGo and its successors, have predominantly relied on model-based Monte-Carlo Tree Search (MCTS). In this work, we present QZero, a novel model-free reinforcement learning algorithm that forgoes search during training and learns a Nash equilibrium policy through self-play and off-policy experience replay. Built upon entropy-
Reinforcement LearningGame AIModel-Free RL
Research arXiv (Computation and Language) Jan 8

What Matters For Safety Alignment?

By Xing Li, Hui-Ling Zhen, Lihao Yin, Xianzhi Yu, Zhenhua Dong, Mingxuan Yuan

82 score
AI Analysis
Comprehensive empirical study on safety alignment evaluating 32 LLMs/LRMs across 13 families using 5 safety datasets, 56 jailbreak techniques, and 4 CoT attack strategies totaling 4.6M API calls.
This paper presents a comprehensive empirical study on the safety alignment capabilities. We evaluate what matters for safety alignment in LLMs and LRMs to provide essential insights for developing more secure and reliable AI systems. We systematically investigate and compare the influence of six critical intrinsic model characteristics and three external attack techniques. Our large-scale evaluation is conducted using 32 recent, popular LLMs and LRMs across thirteen distinct model families, spa
AI SafetyAlignmentJailbreakingLLM Evaluation
Research arXiv (Machine Learning) Jan 8

Why LLMs Aren't Scientists Yet: Lessons from Four Autonomous Research Attempts

By Dhruv Trehan, Paras Chopra

80 score
AI Analysis
Documents four attempts at autonomous ML research paper generation using LLM agent pipeline, with one acceptance at Agents4Science 2025. Identifies six failure modes including bias toward defaults, implementation drift, and weak scientific taste.
We report a case study of four end-to-end attempts to autonomously generate ML research papers using a pipeline of six LLM agents mapped to stages of the scientific workflow. Of these four, three attempts failed during implementation or evaluation. One completed the pipeline and was accepted to Agents4Science 2025, an experimental inaugural venue that required AI systems as first authors, passing both human and multi-AI review. From these attempts, we document six recurring failure modes: bias t
AI AgentsScientific AILanguage ModelsAutonomous Research
Research arXiv (Machine Learning) Jan 8

Jailbreaking LLMs Without Gradients or Priors: Effective and Transferable Attacks

By Zhakshylyk Nurlanov, Frank R. Schmidt, Florian Bernard

80 score
AI Analysis
Introduces RAILS, a black-box jailbreaking framework operating solely on model logits without gradients or handcrafted priors. Matches gradient-based methods through auto-regressive loss for exact prefix matching and history-aware local search.
As Large Language Models (LLMs) are increasingly deployed in safety-critical domains, rigorously evaluating their robustness against adversarial jailbreaks is essential. However, current safety evaluations often overestimate robustness because existing automated attacks are limited by restrictive assumptions. They typically rely on handcrafted priors or require white-box access for gradient propagation. We challenge these constraints by demonstrating that token-level iterative optimization can s
AI SafetyJailbreakingAdversarial AttacksLanguage Models
Research arXiv (Machine Learning) Jan 8

Ratio-Variance Regularized Policy Optimization for Efficient LLM Fine-tuning

By Yu Luo, Shuo Han, Yihan Hu, Dong Li, Jianye Hao

78 score
AI Analysis
Proposes ratio-variance regularization for LLM fine-tuning that constrains policy ratio variance instead of hard clipping, addressing PPO/GRPO's suppression of high-reward divergent actions and improving sample efficiency.
On-policy reinforcement learning (RL), particularly Proximal Policy Optimization (PPO) and Group Relative Policy Optimization (GRPO), has become the dominant paradigm for fine-tuning large language models (LLMs). While policy ratio clipping stabilizes training, this heuristic hard constraint incurs a fundamental cost: it indiscriminately truncates gradients from high-return yet high-divergence actions, suppressing rare but highly informative "eureka moments" in complex reasoning. Moreover, once
Reinforcement LearningLanguage ModelsRLHFOptimization
Research arXiv (Computer Vision) Jan 8

GAMBIT: A Gamified Jailbreak Framework for Multimodal Large Language Models

By Xiangdong Hu, Yangyang Jiang, Qin Hu, Xiaojun Jia

78 score
AI Analysis
Introduces GAMBIT, a gamified jailbreak framework for multimodal LLMs that leverages model reasoning incentives through game-like scenarios. Shows improved effectiveness on reasoning models compared to existing attacks.
Multimodal Large Language Models (MLLMs) have become widely deployed, yet their safety alignment remains fragile under adversarial inputs. Previous work has shown that increasing inference steps can disrupt safety mechanisms and lead MLLMs to generate attacker-desired harmful content. However, most existing attacks focus on increasing the complexity of the modified visual task itself and do not explicitly leverage the model's own reasoning incentives. This leads to them underperforming on reason
AI SafetyJailbreakingMultimodal ModelsAdversarial Attacks
Research arXiv (Artificial Intelligence) Jan 8

STAR-S: Improving Safety Alignment through Self-Taught Reasoning on Safety Rules

By Di Wu, Yanyan Zhao, Xin Lu, Mingzhe Li, Bing Qin

78 score
AI Analysis
Proposes STAR-S, a self-taught framework for improving LLM safety by learning to reason over safety rules. The approach creates a synergistic loop where models elicit reasoning guided by safety rules, then fine-tune to enhance safety reasoning.
Defending against jailbreak attacks is crucial for the safe deployment of Large Language Models (LLMs). Recent research has attempted to improve safety by training models to reason over safety rules before responding. However, a key issue lies in determining what form of safety reasoning effectively defends against jailbreak attacks, which is difficult to explicitly design or directly obtain. To address this, we propose \textbf{STAR-S} (\textbf{S}elf-\textbf{TA}ught \textbf{R}easoning based on \
AI SafetyAlignmentJailbreak DefenseSelf-Improvement
Research arXiv (Computation and Language) Jan 8

RedBench: A Universal Dataset for Comprehensive Red Teaming of Large Language Models

By Quy-Anh Dang, Chris Ngo, Truong-Son Hy

78 score
AI Analysis
Introduces RedBench, a universal red teaming dataset aggregating 37 benchmark datasets with 29,362 samples across 22 risk categories and 19 domains for systematic LLM vulnerability assessment. Provides standardized taxonomy and baselines for modern LLMs.
As large language models (LLMs) become integral to safety-critical applications, ensuring their robustness against adversarial prompts is paramount. However, existing red teaming datasets suffer from inconsistent risk categorizations, limited domain coverage, and outdated evaluations, hindering systematic vulnerability assessments. To address these challenges, we introduce RedBench, a universal dataset aggregating 37 benchmark datasets from leading conferences and repositories, comprising 29,362
AI SafetyRed TeamingBenchmarksLLM Evaluation
Research arXiv (Artificial Intelligence) Jan 8

Controllable LLM Reasoning via Sparse Autoencoder-Based Steering

By Yi Fang, Wenjie Wang, Mingfeng Xue, Boyi Deng, Fengli Xu, Dayiheng Liu, Fuli Feng

77 score
AI Analysis
Uses Sparse Autoencoders to decompose strategy-entangled hidden states in Large Reasoning Models for controllable reasoning. Enables fine-grained control over cognitive strategies like backtracking and cross-verification.
Large Reasoning Models (LRMs) exhibit human-like cognitive reasoning strategies (e.g. backtracking, cross-verification) during reasoning process, which improves their performance on complex tasks. Currently, reasoning strategies are autonomously selected by LRMs themselves. However, such autonomous selection often produces inefficient or even erroneous reasoning paths. To make reasoning more reliable and flexible, it is important to develop methods for controlling reasoning strategies. Existing
InterpretabilitySparse AutoencodersReasoning ControlAI Safety