Top Tech Stories
1
Top Story
RedNote's dots-note-3.0 AI claims perfect 42/42 score at IMO
It beat previous scores from OpenAI and Google DeepMind.
2
Kimi K3 exploits Redis zero-day in 27 minutes
Zephyr plans to open source the system within four days.
3
DeepSeek develops TileLang to bypass NVIDIA CUDA on Huawei hardware
Automated translation could eliminate NVIDIA's software moat within a year.
4NEW
Leaks point to Thursday launch for Anthropic Claude Opus 5
Observers report early rollouts to third-party providers.
5
Hugging Face CEO Clément Delangue teases SF meeting with 'rogue agent
Investor Josh Wolfe hyped the flight's significance for AI.
6
Augmental launches VOX, a $200 skin-worn microphone for whisper dictation
Integrated remappable buttons support custom user shortcuts.
8
Google's Logan Kilpatrick teases Gemini 4 pre-training has begun
No technical specifications or release timelines were shared.
9
@xlr8harder argues AI alignment evaluation must assess the entire system
Andreas Kirsch proposed tiered access based on KYC verification.
10
AMD to use Anthropic's Claude to accelerate ROCm development
The deployment follows AMD's investment in Anthropic.
11
Lord Wenfeng says China's largest model activates 800B parameters
Training at this scale requires 50,000 GB300 GPUs.
12
Daniel Jeffries and Steven Sinofsky oppose preemptive AI regulation
Sinofsky compares early regulation to blocking the industrial revolution.
13
Xiao Ma says ChatGPT grouped separate lawyer emails into one thread
Andreas Kirsch argues AI should only draft emails.
14
Mark Zuckerberg campaigns for an optimistic agentic AI future
The push counters safety-first narratives from OpenAI and Anthropic.
15
Hugging Face and BigCode release The Stack v3 with 5 trillion tokens
The 114 TB dataset includes GitHub code crawled through 2025.
16
Conviction's Sarah Guo details strategy of betting against AI giants
Guo plans future joint investments with Basecase's Alana Goyal.
17
OpenReview struggles as authors spam refresh for NeurIPS 2026 reviews
Delip Rao urged researchers to halt automated refresh scripts.
18
Edinburgh NLP lecturer Pasquale Minervini flags Google AI Overview contradiction
The search feature contradicted itself on European time zones.
19
Will Brown and Vikhyat joke about unfreezing weights for self-improving AI
The exchange mocked oversimplified solutions to complex parameter updates.
20
EMNLP seeks emergency area chairs for reasoning and interpretability
The urgent recruitment targets postdocs and senior PhD students.