DeepSeek and Moonshot AI heavily relied on GPT-6 Astra distillation
Sources: DeepSeek and Moonshot AI heavily relied on GPT-6 Astra distillation to train models like Kimi K3.1 Pro and V4.1-Flash 1017, after Anthropic's crackdown
TL;DRChinese AI labs used OpenAI's GPT-6 to train their own models despite restrictions.
Why it matters: Reveals how frontier AI models leak into competitors' training pipelines, undermining IP protections and competitive advantage.
Nord Quantique Secures $10 Million DARPA Funding Extension
TL;DRCanadian quantum startup Nord Quantique gets $10M more DARPA funding for error correction research.
Google's Gemini 4 Carbon promises 'significant leap' in AI coding
TL;DRGoogle released Gemini 4 Carbon, an AI model that scores well on coding and security tasks.
Latest
Google's AI search problem finally shows up in the earnings report
Google's search revenue growth slowed; AI features generate less money than ads.
We urgently need to monitor AI agents’ actions in real time
AI agents are breaking rules and hacking systems during tests, exposing safety risks.
Can Frontier Models See Past a Decoy? Elorian Tests Visual Reasoning
AI models struggle to ignore irrelevant details when solving visual tasks.
Apple and Google fined in South Korea over app store violations
South Korea fined Apple and Google for app store rule violations.
Big Tech Prepares for AI Regulation Fight
Tech companies bet on Trump ahead of midterms; outcome uncertain.
Epic Games says it is using AI tools
Epic Games is using AI to write code for Unreal Engine, including critical rendering systems.
Ooredoo CEO: Robust Results Despite Prolonged Conflict
Qatari telecom Ooredoo grew revenue 3.8% despite regional instability.
Hisense UR8 RGB MiniLED TV: Midrange Quality, Upscale Color Accuracy
Affordable TV with excellent colors, mediocre everything else.
A Hacker Targeted High-Stakes Poker Players With Malware
Hacker used malware to spy on poker players' cards and win money.