synthetic data generation
Synthetic data generation means creating training examples by algorithm rather than collecting them from the real world. Instead of hiring people to write questions or scraping websites, a system produces its own text, tasks, or environments, often starting from a small seed or a structured rule. The goal is to get abundant, controllable, and sometimes verifiable data for training AI models, especially when natural data is scarce, expensive, or privacy-sensitive. The quality of this data depends on how it is generated, filtered, and checked for correctness.
Across these papers, the common thread is that synthetic data works best when it is tightly coupled to the model's own learning process. Some extract tasks from existing code or executable environments, others have a model rewrite or translate its own data, and several use self-play or adaptive curricula so the model keeps pushing just beyond its current ability. They also emphasize validation: gates, execution feedback, and quality checks separate useful examples from misleading ones. Together, they show synthetic data is moving from a stopgap to a deliberate, scalable design choice.
- Papers
- 97
- Released code
- 3
- First seen
- Sept 2026
- Latest
- Nov 2026
97 papers in the last two months, against 0 in the two before.
Start here
Three papers to read first, in the order they came out- 1AI trained on simulated data can reconstruct unfamiliar medical scans
Training on physics-simulated synthetic data unlocks scaling laws for zero-shot MRI reconstruction.
Nov 2026 - 2Coding agents learn more when training projects are revised automatically
RepoFactory converts real repositories into training environments where coding agents build complete projects, improving benchmark performance by up to 22.6 percentage points.
Nov 2026 - 3AI agents can learn cross-service workflows with fewer training examples
CrossEnv generates verifiable cross-environment tasks whose curriculum delivers strong performance with fewer examples.
Nov 2026
How fast it is moving
Often studied with
verificationChecking outputs for correctness before accepting them, often to catch errors or reward hacking.18 togetherbenchmarkA standard set of tasks or data used to evaluate and compare the performance of AI systems.15 togetherlarge language modelA neural network trained on massive text corpora to generate coherent text and perform diverse language tasks.11 togetherWho is working on it
The papers
Most central to this idea first, not most recent.
- Top Universitycs.CL
Coding agents learn more when training projects are revised automatically
RepoFactory converts real repositories into training environments where coding agents build complete projects, improving benchmark performance by up to 22.6 percentage points.
Beihang University, Shanghai Jiao Tong University · Nov 2026
- Top Universityeess.IV
AI trained on simulated data can reconstruct unfamiliar medical scans
Training on physics-simulated synthetic data unlocks scaling laws for zero-shot MRI reconstruction.
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen University · Nov 2026
- Chinese Techcs.AI
AI agents can learn cross-service workflows with fewer training examples
CrossEnv generates verifiable cross-environment tasks whose curriculum delivers strong performance with fewer examples.
Beijing Institute of Technology, Alibaba Group · Nov 2026
- Big Techcs.AI
Simulated business histories yield verifiable benchmarks that separate AI agent performance
SynthCorp simulates business events to generate consistent data estates and verified benchmarks for data agents.
Google · Nov 2026
- Industrycs.AI
Language models nearly match peers using one-seventh the active parameters
K2-Horizon releases six fully open language models, including a 375B MoE, with leading agentic and coding performance.
Institute of Foundation Models · Nov 2026
- Top Universitycs.LG
Generalist tabular models match specialists on utility, not fidelity
Foundation models show strong performance across generation, imputation, and prediction on tabular data.
StableAI, Tsinghua University · Nov 2026
- Big Techcs.LG
AI models trained on artificial tables lead eight prediction benchmarks
Kumo Tabular foundation models achieve state-of-the-art accuracy on eight tabular prediction benchmarks while running faster than competitors.
NVIDIA, INRIA · Nov 2026
- Big Techcs.AI
AI agents nearly triple success rate in solving crystal structures from powder X-ray diffraction
LLM agents with verifier-guided adaptation recover 69% of crystal structures from powder X-ray diffraction, up from 27%.
Stanford University, University of Toronto · Nov 2026
- Top Universityeess.AS
229K-hour dataset lets speech models generate minutes-long audio with control
Emilia2 is a 229K-hour dataset for controllable long-form speech generation across 41 languages.
The Chinese University of Hong Kong, Shenzhen, University of Illinois at Urbana-Champaign · Nov 2026
- Independentcs.AI
AI agent benchmarks add work but not depth
Most synthetic long-horizon terminal tasks add operations, not required sequential turns, and depth better predicts difficulty.
Nov 2026
- Big Techcs.RO
Humanoid robots trained in simulation can navigate homes without prior mapping
A humanoid robot trained only in simulation achieves 91% success performing physical tasks in unfamiliar real homes.
Apple · Nov 2026
- Big Techcs.LG
AI forecasting model can rival specialist tools without dataset-specific training
Chronos-2 forecasts interacting time series and covariates in zero-shot, beating specialized models in accuracy and speed across domains.
Amazon Web Services, University of Freiburg · Nov 2026
- Big Techcs.LG
AI agents learn more from varied tools than extra tasks
To scale tool-use agent training data, grow tool and database diversity, not just task count.
Apple · Oct 2026
- Chinese Techcs.AI
Small language models judge outputs more accurately when trained on counterfactual pairs
Counterfactual pairwise augmentation (CPA) improves pointwise judges by training on label-flipping response pairs.
Huawei Technologies Ltd. · Nov 2026
- Chinese Techcs.LG
A frozen AI model can imitate an unseen dataset from one example
A pretrained generator can create realistic new time series for an unseen domain from a single reference sequence, without retraining, by reusing learned shape and shape-transition vocabularies.
State Key Laboratory of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences, University of Chinese Academy of Sciences · Nov 2026
- Top Universitycs.LG
Scaling laws predict when synthetic rewrites improve language model training
The optimal fraction of synthetic rewrites in pretraining grows with compute and is predictable from a scaling law, and rewritten models can beat their rewriter.
Princeton University, Stanford University · Oct 2026
- Chinese Techcs.AI
AI agents get nearly ten times more training tasks from scientific code
Poseidon generates nearly ten times more training environments from scientific code repositories than prior methods and improves agent performance.
The Hong Kong University of Science and Technology (Guangzhou), Hunyuan Team, Tencent · Oct 2026
- Big Techcs.CL
Synthetic translated data improves language models across twelve Indian languages
A 2.83-trillion-token synthetic corpus plus quality-controlled evaluation improves Indic-language knowledge and reasoning over open-corpus training without measurably reducing English performance.
NVIDIA · Oct 2026
- Big Techcs.CR
Guided diffusion enables trustworthy synthetic text with formal privacy guarantees
Guided Diffusion steers diffusion language models toward target semantic embeddings to generate synthetic text with differential privacy and high utility.
NVIDIA, Carnegie Mellon University · Oct 2026
- Chinese Techcs.CL
Language models can better balance safety and usefulness with less data
Beijing Normal University, Tencent · Nov 2026
- AI Startupcs.CL
Language models gain no advantage from synthetic warmups at equal compute
Synthetic pre-pretraining never outperforms natural-language training when compute is matched across tested regimes.
Cohere, AI Centre, UCL · Oct 2026
- Big Techcs.CL
AI chatbots' psychological risk factors can be monitored without measuring user harm
Detecting eight behavioral risk factors in LLM responses enables scalable psychological harm monitoring without measuring harm directly.
Microsoft · Nov 2026
- Big Techcs.CR
AI agents cut prompt injection success from 79% to 9%
SecPO combines a new training objective with strong simulated attacks to cut prompt-injection success while preserving performance on benign tasks.
UC Berkeley, Anthropic · Nov 2026
- Independentcs.CV
Video generators can use simulator checks to correct physical errors
PhysLoop improves video generation models by using a simulator to guide an LLM's physics knowledge into generating physically consistent training videos.
Nov 2026
- Academiccs.CV
Practicing reusable web skills then composing them improves compact agents
Training web agents on reusable interaction skills, then decomposing tasks into those skills, improves success on unseen websites and external benchmarks.
University of Utah, Islamic University of Technology · Oct 2026
- Big Techcs.CL
AI agents nearly clear proof benchmarks but harder code challenges remain
EasyProver proves almost all existing formal math and code benchmarks and introduces a harder benchmark from competitive programming.
Case Western Reserve University, Amazon Web Services · Oct 2026
- Independentcs.LG
Training on iterative repair data lifts AI GPU kernel correctness from 0% to 82%
KernelMosaic dataset and agentic framework raise CUDA kernel generation correctness from near zero to 88%, 82%, and 16% on KernelBench L1, L2, L3.
Nov 2026
- Top Universitycs.AIcode
Reasoning models transform operations, not concepts in puzzles
Large reasoning models form stable abstractions of transformation operations that generalize across diverse ARC-style tasks.
LMU Munich, MCML · Nov 2026
- Big Techcs.SD
AI agents locate sounds better after training with written reasoning
A spatial audio agent answers questions, holds dialogue and calls tools on four-channel ambisonic recordings, scoring 44.9 percent on a new 12,000-sample benchmark, 7.4 points above prior systems.
Microsoft Research · Nov 2026
- Big Techcs.CV
AI models struggle to follow events across directions in panoramic video
Panoramic video reasoning models can improve by 12.5 points through sphere-time grounded reasoning training.
Nanjing University, NVIDIA · Nov 2026
- Top Universitycs.AI
Backtesting method lets AI agents forecast past events for instant feedback
FABench reveals retrieval helps agents match market forecasts, yet evidence weighing remains a critical failing.
Raven Labs, Zhejiang University · Nov 2026
- Big Techcs.AI
AI model estimates treatment effects faster without retraining for each dataset
PICLE-PFN is a foundation model for longitudinal causal effect estimation that supports dynamic policies and predicts outcomes in one forward pass without retraining.
Cornell University, University of Washington · Oct 2026
- Chinese Techcs.CV
Image editor handles 3D pose and identity together from one instruction
DecoEdit performs unified 3D-aware image editing by decoupling geometry from detail preservation, surpassing proprietary models.
Zhejiang University, Alibaba Group · Nov 2026
- Top Universitycs.LG
Language models can improve coding answers without checking target correctness
Test-Time Self-Play beats verifier-based methods for test-time scaling by training a solver on generated synthetic problems, improving by 35% with no verifier.
ETH Zurich, Tufa Labs · Nov 2026
- Independentcs.AI
A small AI agent trained on cloned websites matches bigger web agents
A suite of 153 self-hosted replica websites with 3,518 tasks trains small web agents that then handle unseen tasks and match leading open agents.
Nov 2026
- Chinese Techcs.CV
AI models struggle to follow television dramas lasting tens of hours
Current video-question-answering models struggle to answer questions about events across entire television dramas, exposing the need for better long-term video memory.
Huawei Technologies Ltd · Nov 2026
- Chinese Techcs.MA
AI agents complete only one in 16 collaboration tasks
A benchmark built from recombined collaboration structures tests whether multi-agent AI systems really coordinate, and none of the tested setups solved more than 6.25% of its tasks.
Xiamen University, ByteDance · Nov 2026
- Big Techcs.CL
Language models improve using self-generated questions and answers, without human examples
Large language models self-improve using only self-generated questions and privileged reasoning, without human-curated data.
Google, Ulm University · Oct 2026
- Chinese Techcs.CV
Image generators help object detectors recognize rare categories more accurately
Tail-SD, a hybrid layout-instruction diffusion model, improves long-tail detection, achieving 4.89% rare box AP and 5.68% rare mask AP on LVIS.
University of Auckland, Dcar · Nov 2026
- Top Universitycs.CV
Video generators better preserve subjects’ appearance as cameras circle them
OmniShowcase generates subject-consistent videos from multi-view images using geometry-aware conditioning and automated data synthesis, achieving state-of-the-art identity preservation and 3D consistency.
FLIPTION, KAIST AI · Oct 2026
- Independentcs.CL
Small AI agents can handle computer tasks with clicks and code
Mage-VL-Cowork, a 4B hybrid computer-use agent, achieves 66.0% on OSWorld-Verified via evolved harness and five-stage training.
Nov 2026
- Big Techcs.AI
Synthetic computer worlds improve agents on unseen computer-use tasks
WildAWM uses open-source models to synthesize computer-use environments spanning GUI, MCP, and CLI; training only in these worlds improves agents across eight out-of-distribution benchmarks.
UNC-Chapel Hill, Microsoft Research · Oct 2026
- Big Techcs.AI
Graded curricula extracted from an environment's latent rules substantially improve trained agents
Gradus composes guarded behavioral operators into calibrated curricula, lifting a 30B model from 59% to 89% on tau2-Bench and 47% to 83% on CAR-Bench.
AWS AI, University of Southern California · Oct 2026
- Top Universitycs.CL
LongZero trains LLMs on self-written long-context tasks, no external data
LongZero improves LLM long-context reasoning via self-generated training tasks without any external data.
Tsinghua University, Tsinghua-Bosch Joint ML Center · Oct 2026
- Big Techcs.CL
AI agents break chatbot safeguards more often in fewer exchanges
An autonomous multi-turn jailbreak attacker can adapt to victim replies, decide when to stop, and achieve SOTA success rates without an in-loop judge.
University of Rochester, Microsoft Research · Oct 2026
- Big Techcs.CV
One model generates matched video and motion for arbitrary rigs
JCA co-generates scene-aligned video, camera, and replayable 3D motion for arbitrary rigs more faithfully than evaluated video-first pipelines.
NVIDIA, Technion – Israel Institute of Technology · Oct 2026
- Big Techcs.CR
Privacy-aware rewriting offers best privacy-capability trade-off for LLMs on sensitive data
Privacy-aware rewriting of synthetic data achieves a more favorable privacy-capability trade-off than direct finetuning or DP-SGD.
Johns Hopkins University, OpenAI · Oct 2026
- Independentcs.CV
AI reconstructs hand movements more accurately from single-camera, first-person video
EgoHand reconstructs 3D hands from egocentric video with state-of-the-art accuracy and low latency using a model trained on 14 million diverse images.
Oct 2026
- Big Techcs.CV
3D diffusion model generates body CT masks with controllable anatomy, tumors, demographics
A 3D latent diffusion model generates whole-body CT masks with joint control over organ size, tumors, and demographics.
NVIDIA, National Institutes of Health · Oct 2026
- Big Techcs.SD
Dense perceptual training improves reasoning in large audio-language models
Constructing post-training datasets with multi-aspect perceptual grounding and cognitive refinement improves audio reasoning accuracy in large audio-language models.
Carnegie Mellon University, NVIDIA · Oct 2026
- Big Techcs.CRcode
New methods tighten privacy accounting for DP synthetic data pipelines
GDP and PLD analyses of bounded-range mechanisms enable tighter end-to-end accounting for mixed DP synthetic data pipelines, improving utility at the same privacy level.
UCL, SAS · Oct 2026
- Big Techcs.LG
A synthetic-trained model predicts across tabular datasets without retraining
TabFM is a zero-shot foundation model for tabular data that outperforms tuned AutoML pipelines across 51 benchmarks.
Google Research · Oct 2026
- Top Universitycs.AI
Self-play on generated programs produces transferable language-model capabilities
Self-play pretraining with synthetic data from a universal Turing machine yields predictable zero-shot scaling on natural data.
Independent Researcher, Tel Aviv University · Sept 2026
- Big Techcs.CV
Language models trained only on text can partly locate video events
A language model trained only on text descriptions can localize events in video without any video training, if frames are mapped to their nearest text embeddings first.
Amazon, University of Oxford · Nov 2026
- Big Techcs.CL
Agent Skills teach AI systems precise subagent permissions
Procedural knowledge in Agent Skills improves subagent authorization precision beyond what task outcomes alone provide.
Samsung Research, Peking University · Nov 2026
- Independentcs.AI
AI forecasting models take the top two spots across three benchmarks
Timestral's four time series foundation models set new state-of-the-art accuracy on multiple forecasting benchmarks.
Oct 2026
- Research Labcs.CV
Vision inputs expose greater social bias than text in matched tests
Switching from text to vision reduces fairness in multimodal LLMs by 0.291, a gap not due to poor visual comprehension.
Technical University of Munich, Helmholtz Munich · Oct 2026
- Chinese Techcs.CV
AI model controls characters in unfamiliar game worlds in real time
An agentic game world model that generalizes zero-shot to novel scenes, characters, and motions with real-time interaction.
Tencent, National University of Singapore · Oct 2026
- Top Universitycs.CV
One model edits people, voices and movements in sync
AVE unifies mask-free human-centric audio-visual editing for person swap, insert, delete with superior synchronization.
University of Hong Kong, National University of Defense Technology · Oct 2026
- Big Techcs.SD
Audio training teaches one model to edit scenes across modalities
Training on synthetic audio edits and masked audiovisual reconstruction enables zero-shot, instruction-driven editing of movie scenes across image, audio, and video modalities.
Adobe Research, Carnegie Mellon University · Oct 2026
- Top Universitycs.CL
Full-duplex speech model scaled to long, multi-party bilingual conversations
A single end-to-end speech model sustains coherent multi-party bilingual conversations over extended durations and outperforms existing open-source models on MultiTalkBench.
CUHK MMLab, CPII under InnoHK · Oct 2026
- Top Universitycs.CR
Adaptive testing exposes major security repair gaps in coding agents
SecProbe adaptively evaluates coding agents on cybersecurity vulnerabilities, requiring up to 29.5% fewer tasks while maintaining comparable ability estimates.
University of Notre Dame, Bake AI · Sept 2026
- Chinese Techcs.CV
Post-training helps an agent sustain complex work for over 10 hours
Marathoner achieves ultra-long-horizon autonomous execution with 1000+ tool calls over 10+ hours, beating proprietary models.
FIC, University of Macau, Ant Group · Sept 2026
- Top Universitycs.AI
Language models learn categories poorly, especially flexible context use
LLMs learn rigid category associations from verbatim repetition but more flexible knowledge from diverse expressions in pretraining data.
Carnegie Mellon University, Bar-Ilan University · Nov 2026
- Research Labcs.RO
Robots learn from human videos by preserving contact with objects
Ego2X converts egocentric human video into robot data by focusing on object contacts, improving physical contact accuracy and downstream policy success.
Beijing University of Posts and Telecommunications, Institute of Automation, Chinese Academy of Sciences · Nov 2026
- Top Universitycs.AI
AI models can read pictured problems yet fail to solve them
Multimodal LLMs fail to reason on images they can read because a reasoning-access gap blocks visual input from using text reasoning machinery.
National University of Singapore, CNRS@CREATE · Nov 2026
- Academiccs.LG
AI forecasters can reason themselves into logical consistency
Using a finetuned language model to reconcile reasoning improves both logical consistency and accuracy of LLM forecasts.
Bridgewater AIA Labs, Massachusetts Institute of Technology · Nov 2026
- Big Techcs.AI
Language models improve psychiatric interviews after training with simulated patients
Training a language model with multi-turn reinforcement learning in simulated psychiatric intakes improves its diagnostic, communication, and safety performance over its base model.
Google DeepMind, Google Research · Oct 2026
- Independentcs.CL
AI models predict training data side effects better than humans
Automated search can generate ordinary-looking training datasets that trigger unexpected language-model behaviors beyond the data’s apparent purpose.
Nov 2026
- Big Techcs.CR
AI agents remain vulnerable to prompt injection when attackers use collaborating swarms
Prompt injection attacks remain effective against frontier AI agents when attackers use swarms of collaborating LLMs.
Google DeepMind · Nov 2026
- Big Techcs.CV
AI image editors get better results comparing alternatives than repeatedly refining
Image editing agents fail because they lack diverse tools and their judges lack quantitative evidence; EditAgent solves both.
Adobe Research, UC San Diego · Oct 2026
- Chinese Techcs.CV
Training with deceptive visual traps makes AI agents correct their own perception errors
Training vision-language models with deliberately misleading visual descriptions, filtered so they change the answer, teaches them to self-correct errors and improves reasoning.
Peking University, ByteDance · Nov 2026
- Big Techcs.AI
Chatbots follow higher-priority rules better after training on conflicting goals
Training on goal-goal conflicts significantly improves language models' adherence to instruction hierarchy.
NVIDIA, UC Berkeley · Oct 2026
- Chinese Techcs.RO
Robots learn tasks better with demonstrations generated by a video model
Shared video-derived latent actions let one world model learn controllable visual dynamics across humans, robots, and simulations without matching their native control spaces.
Hong Kong Embodied AI Lab, CUHK · Oct 2026
- Independentcs.SD
Voice model matches text reasoning without separate transcription step
Apriel Voice 27B Preview is a single open model that understands speech, reasons, converses, and calls tools.
Nov 2026
- Big Techcs.SD
Audio model composes timestamped captions for queryable timelines
TAC2 constructs granular, speaker-attributed acoustic timelines using composable timestamped audio captioning functions.
University of Surrey, Adobe Research · Oct 2026
- Chinese Techcs.AI
Language models struggle to turn social understanding into appropriate responses
SocialArena systematically evaluates LLM social intelligence via role-playing dialogue and human judgment.
Alibaba Token Hub, Alibaba Group · Oct 2026
- Chinese Techcs.SE
Token-level evidence pinpoints AI-written lines and blocks inside mixed code files
Token-level likelihood ratios plus contextual line and block aggregation detect and localize small AI-involved code spans better than file-wide scoring.
Peking University, Baidu Inc. · Oct 2026
- Academiccs.AI
New benchmark reveals agents fail to detect and recover from their own errors in long tasks
Computer-use agents rarely recover from their own errors in long-horizon workflows; fine-tuning on real failures improves recovery and clean-start performance.
University of Texas at Dallas, Independent Researcher · Oct 2026
- Industrycs.LG
AI agents get better at software tasks by reusing their own verified work
Generating software tasks, recording how agents execute them, and retraining on verified runs produces open-source computer-use agents that lead their size class on screen, desktop and web tests.
Shanghai Qiji Zhifeng Co., Ltd. · Nov 2026
- Chinese Techcs.LG
Financial multi-hop benchmark exposes sharp agent performance drop beyond four hops
FinHopper is a 1-6 hop financial retrieval benchmark with gold intermediate answers showing agent performance drops sharply with depth.
South China University of Technology, FiT, Tencent · Oct 2026
- Chinese Techcs.LG
Evidence annotations improve step-level credit for deep search agents
Evidence-guided step-level process rewards derived from annotated evidence claims improve credit assignment and outperform standard GRPO on deep search benchmarks.
Renmin University of China, Key Laboratory of Data Engineering and Knowledge Engineering, MOE · Oct 2026
- Big Techcs.CV
Benchmark tests multi-camera 3D tracking across synthetic and real spaces
NVIDIA's Physical AI Smart Spaces benchmark provides 280+ hours of multi-camera 3D perception data for warehouses and smart spaces.
NVIDIA, Santa Clara University · Oct 2026
- Research Labcs.AI
Active visual probes help browser agents act more accurately
Active visual probes improve browser-agent task success by verifying DOM-to-pixel alignment on demand while retaining compact, decision-relevant memory.
Institute of Computing Technology, Chinese Academy of Sciences, University of Chinese Academy of Sciences · Sept 2026
- Top Universitycs.CVcode
Fine-Grained Training Data Improves Agents on Interactive CAPTCHA Tasks
CaptchaArena supplies 50,000 execution-verified interactive CAPTCHA puzzles across 20 types, and CaptchaAgent trained on it reaches 71.7 Pass@1.
Columbia University, Zhejiang University · Sept 2026
- Research Labcs.CV
Visual agents improve by jointly evolving reusable skills and practice data
V-Gym co-evolves procedural skills and multimodal practice data from agent execution trajectories for continual self-improvement.
State Key Laboratory of AI Safety, Institute of Computing TechnologyChinese Academy of Sciences · Sept 2026
- Big Techcs.CL
Voice-generating AI models produce far less diverse speakers than real humans
Instruction-guided text-to-speech systems produce far less diverse voices than real speech, especially for female prompts.
National Taiwan University, NVIDIA · Nov 2026
- Chinese Techcs.CL
AI agents fail nearly a quarter of office tasks at best
Agent pass rates on single-domain enterprise tasks range from 22.6% to 76.4% depending on model and interface.
Alibaba Group, HKUST(GZ) · Nov 2026
- Top Universitycs.CV
Joint video and 3D motion prediction reduces trajectory errors by over half
WorldFlow jointly predicts future video and queried three-dimensional object motion, achieving the best displacement errors among evaluated learning-based methods on three test sets.
RoboScience, National University of Singapore · Oct 2026
- Chinese Techcs.AI
AI agents can match frontier models at training other models
Training LLM agents on RSI-Dojo trajectories improves PostTrainBench scores by 8.74 to 16.87 points and matches frontier model performance.
The Chinese University of Hong Kong, ByteDance · Nov 2026
- Big Techcs.CL
Rewarding moral arguments makes a language model more consistent, not sounder
Rewarding argument quality makes language models more decisive and order-invariant on moral dilemmas without improving moral validity, so consistency is not validity.
Google DeepMind · Oct 2026
- Big Techcs.CV
Video AI answers 3D questions better when it predicts how objects move
Teaching a video vision-language model to write out ordered 4D object states and reason over how they change between timesteps improves its answers to spatiotemporal questions.
KAIST AI, Microsoft Research Asia · Nov 2026
- Chinese Techcs.AI
AI agents improve when training tasks adapt to their weaknesses
EvoMCP improves AI agents by co-evolving training tasks and demonstrations based on student feedback.
Shanghai Jiao Tong University, Ant Group · Nov 2026
- Chinese Techcs.CV
Verifiable rewards help video captioning models capture more complete visual narratives
MVRCap frames video captioning as reinforcement learning with hierarchical multi-dimensional verifiable rewards, achieving state-of-the-art results on VDC, CAPability, and VC4VG.
Xiamen University, Alibaba Group · Oct 2026
- Chinese Techcs.AI
Large-scale dataset and world-model RL improve embodied navigation in open worlds
Embodied world agents improve open-world navigation-planning via offline pretraining on a 200K-sample dataset and interactive reinforcement-learning post-training inside a video world model.
Tencent Hunyuan · Oct 2026
- Academiccs.LG
Generating frontier-level problems improves reinforcement learning for LLM reasoning
LLM reasoners improve by continually generating training problems at the edge of their evolving capability rather than using a fixed problem pool.
University of Basel, University College London · Sept 2026
- Chinese Techcs.AI
PlanGuard detects physical risks across complete embodied-agent plans
PlanGuard detects physical safety risks in complete multi-step embodied plans before execution, reaching 87.15% accuracy and 87.21% F1 at 2B scale.
Anhui Province Key Laboratory of Digital Security, The University of Hong Kong · Sept 2026
Concepts are extracted from each paper and reused across the corpus, so this page grows on its own as the desk reads.