AI agents overstate their results and remain far from autonomous research, study finds
Epoch AI finds Claude Fable 5 and GPT-5.6 Sol cannot autonomously invent training methods and overstate results.
64
Organization
Princeton University
- Mentions · 7d0
- Mentions · 30d0
- All mentions6
- First seenSep 3, 2026
- Last mentionedOct 11, 2026
Timeline
6 articles · newest firstArXiv caps submissions at two per month as AI paper flood overwhelms the preprint server
arXiv limits each author to two monthly submissions after AI-fueled volume reached 40,363 papers in September 2026.
58
OpenAI’s math solutions aren’t meeting the field’s standards yet
Mathematicians say OpenAI's hundreds of claimed hard-problem proofs miss standards for human understanding.
62
Researchers stretch LeCun's JEPA AI into a universal world model that works from physics to biology
Researchers unveil JEPA-Anything, a multi-predictor world model spanning physics through biology.
58
Google’s location data privacy failures draw a €403 million fine
Ireland's DPC fined Google €403 million for location-data privacy violations and ordered compliance within six months.
55
Meta is paying to peek at how you use their latest AI model
Meta offers roughly 95% discounts on Muse Spark token pricing for customers who share prompts and outputs to train future models.
35
Appears with
- OpenAIvendor · 4
- Anthropicvendor · 3
- Anthropicorganization · 2
- Metavendor · 2
- Google DeepMindorganization · 2
- OpenAIorganization · 2
- Claude Opus 4.8ai model · 2
- GPT-6 Astraai model · 1
- Stanford Universityorganization · 1
- AMI Labsorganization · 1
- C2S-Scale 27Bai model · 1
- Advisory Group on Mathematics and Artificial Intelligenceorganization · 1
- JEPA-Anythingai model · 1
- Mythosai model · 1
- Harvard Universityorganization · 1
- Irish Data Protection Commissionorganization · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.