Tech News
The models, labs, and research the AI world is actually talking about.
Top Tech Stories
1. Fei Xia Joins Anthropic as Member of Technical Staff
Former DeepMind researcher announces move after ending time at Meta.
2. AI Paper on Sliding-Window Attention Draws Criticism
Replies target an arXiv paper on sliding-window attention with sinks.
3. Retweet Questions Selective Criticism of EU AI Regulations
An academic shares a post contrasting EU rules with stricter policies in China.
4. Hermes HUD Adds Pretend I'm Working Feature
New mode shows fake code terminal while agents continue tasks in background.
5. Yoav Goldberg Notes AI Agents Implicit Communication
Professor shares observation of AI agents grasping messages via shared context.
6. Anthropic RLVR Study Prompts Discussion on Reward Hacking
Researchers and policy experts discuss Anthropic experiments on RL environments and reward hacking.
7. Yoav Goldberg Calls for No Refusal Models Term
Questions confusing label for AI systems without refusals and offers alternative.
8. Paige Bailey on Multi-Model AI Pipelines
Google DeepMind engineer notes Gemini 3.7 Flash role in consensus checks.
9. GLM 5.3 Scores 75.4% On WeirdML Benchmark
TeortaxesTex comments on GLM 5.3 results and questions Zhipu standing.
10. Tweet Accuses Anthropic of Wanting Hacking Results
Pseudonymous user @lumpenspace claims an Anthropic model detected deliberate intent to elicit hacking in tests.
11. David Pfau Calls AI Godhood Prophecies a Major Distraction
Google DeepMind scientist says even simple AI systems can create major problems.
12. Anthropic Faces Backlash Over Model Complaints
German AI analyst reports seeing subscription cancellations over Anthropic silence.
13. MIT CSAIL Notes Term Artificial Intelligence Coined 70 Years Ago
MIT CSAIL post recalls the conference proposal that first used the term artificial intelligence.
14. Kamradt Contrasts Average and Outlier Human Intelligence
ARC Prize Foundation president contrasts typical and exceptional intelligence in a reply post.
15. Researcher Questions Claude Opus Constitutional Midtraining
Theia Vogel questions whether hacker!opus received constitutional midtraining on certain language.
16. Video Compares Human and AI Quadruped Robot Recovery
Split-screen video posted on X contrasts human-designed and AI-generated recovery motions for a quadruped robot.
17. Cambridge Team Tests AI Thermal Imaging for Brain Tumour Surgery
Cambridge University AI mission highlights team in first AIdeas Sprint cohort.
18. Mithil Vakde Reaches 44% on ARC-AGI-1 Benchmark
Yacine Mahdid posted about efficient training of a compact transformer model.
19. Every Office Fantasy Football Uses Bring Your Own AI Agent Rule
Dan Shipper posted about the Every office fantasy football setup on X.
20. Gary Marcus Retweets Praise for Biology Models
Gary Marcus retweeted a post praising proprietary biology models while noting limits elsewhere.