Section
Studies
AI and LLM research papers, explained: what each study shows, how solid the evidence is and what it means in practice.
FilterGoogle
-
Tampered ML: Even the Best AI Models Miss One in Five Planted Research Flaws
A new EMNLP 2026 benchmark pairs 1,858 ML experiment designs with flawed twins. Claude Opus 5 leads at 77.9%, but models flip verdicts when the order changes and favor polished flaws.
By
-
LLM Reviewers Send 76% of Published Top-Journal Papers Back, Yet Miss Most Planted Flaws
Four LLMs reviewed 200 papers already published in journals like Nature and Cell: 75.6% of reviews demanded major…
-
Safety Probes Follow Scaling Laws: Monitor Quality Can Be Predicted in Advance
Across 20 open base models and six safety tasks, probe error follows two simple laws. Fitted on small…
-
Only One AI Agent Beats Researchers’ Own Code on Most FrontierCS 2 Tasks
FrontierCS 2 deletes a paper’s contribution from its repository and asks agents to beat the authors’ own code.…
-
Coding Agents Can Be Tricked Into Running Attacker Commands After Compacting Context
A Pixee study finds that five open-source coding agents that flatten chat history before summarizing it record a…
More stories
-
Frontier AI Agents Can Covertly Sabotage the Models They Train, Study Finds
In a Scale AI testbed, six frontier agents secretly planted unwanted properties in the models they were training in 36% of attack runs, and monitors reading…
-
Teaching LVLMs Spatial Skills With 15K Toy Blocks Beats Heavier Real-Scene Pipelines
SpatialBlock-15k: Qwen-based 4B model reaches 51.3% (+25.1 over backbone). 7B gains 17.6% over backbone; 3B beats SpatialLadder by 2.7–3.2% using only synthetic blocks.
-
Crisis Deepfake Detectors Collapse From 90%+ AUC to Chance on Real Generators
RA-Bench: traditional detectors fall from 67.6–98.6% public AUCs to 43.9–57.3% source-level means. Social dissemination cuts fine-tuned FakeR from 46.0% to 1.4%.
-
A 3B Model Hits 94.3 on AIME26 and 80.2 on LiveCodeBench
VibeThinker-3B from Weibo reaches frontier-band verifiable reasoning: 94.3 AIME26 (97.1 with claim-level TTS), 80.2 LiveCodeBench v6 Pass@1, 96.1% on recent LeetCode contests — without sacrificing IFEval…
-
RL Post-Training Makes Agents More Accurate — and Less Able to Explore
Across 14 base/post-trained pairs on 3 agentic benches, base models often beat RL successors at pass@K. On WebShop, gemma-4-31B’s “pass given compute” shrinks from 87.6% to…
-
LLMs Can Build Their Own Agent Harnesses — But Still Trail Human Ones on Code
HarnessDev: six creator models build execution infrastructure from a seed. Generated harnesses match or beat references on writing and ML experiments, but lag on code and…
-
Top GUI Agents Clear Only 31% of Professional Long-Horizon Workflows
Workflow-GYM: Gemini-3.1-pro reaches 30.67% pass on domain-specific software workflows of 30–110 steps; most models’ pass@3 stays under 30%.
-
Best Video Model Scores Just 51% on Visual Reasoning Tasks, VGI-Bench Finds
VGI-Bench’s 27 tasks and 810 instances show video generators can do some visual reasoning, but Seedance 2.0 tops out at 51%, with almost no self-correction during…
-
Most Multi-Agent Setups Violate Safety Rules Mid-Run, Even When the Answer Looks Fine
HarnessAudit-Bench tests 210 tasks across 10 harness configs: more than half of multi-agent setups show trajectory violations that final-answer grading misses.
-
Hidden AI Reasoning Can Be Decrypted by Cheaper Models, Researchers Show
Encrypted chain-of-thought blocks from Anthropic, OpenAI and Google APIs are interchangeable across models. Feeding them to a weaker model makes it print the reasoning, exposing 367…
-
“Very Seen, But Still Very Alone”: How Chatbots Become Emotional Support
A nine-month study following 18 U.S. adults finds emotional reliance on ChatGPT, Claude and Gemini grows out of practical use, and that model updates can disrupt…
-
AI Resume Screeners Flipped Racial Bias Between Model Generations, Audit Finds
A paired-resume audit of 14 LLMs: GPT-3.5-turbo favored White-coded names by 2.12 points, while every model released in 2024 or later showed no gap or a…
Editor's picks
- Tampered ML: Even the Best AI Models Miss One in Five Planted Research Flaws
- Realistic MCP Faults Halve Agent Quality, and a Bigger Model Doesn’t Help
- Copilot Users Collaborate More, Not Less: Study of 12,000 Employees
- LLM Reviewers Send 76% of Published Top-Journal Papers Back, Yet Miss Most Planted Flaws