AI Daily Brief: USAF AI Agent Flies X-62A; Hassabis Exits DeepMind?
Audio in Mandarin Chinese · English transcript below
⚡ AI F-16: 27 live intercepts; DeepMind exodus; China AI accel; slim LLMs vs regs.
The AI industry is turning away from indiscriminate expansion and toward sharper focus and strategic retrenchment. Kimi K3’s decision to drop multilingual support cut its model from 711GB to 478GB—a deliberate trade-off of non-core capabilities for extreme leanness. Alphabet, meanwhile, reorganized DeepMind’s leadership, shifting the organization from front-line work to strategic positioning and allowing Demis Hassabis to concentrate on research. Together, the two moves confirm that subtraction, not sheer scale, has become the industry’s consensus. Notably, DeepMind’s strategic pivot arrives amid reports of core talent attrition, a reminder that even a focused realignment comes with organizational growing pains.
Today's Top 3 Headlines
- AI Industry News
🤖 AI-piloted F-16 completes 27 live-fire intercepts, India pushes AI labeling policy, DeepMind restructures
USAF & Lockheed Martin revealed AI-piloted X-62A completed 27 live-fire intercepts; Stanford & Arc Institute used DNA LLM Evo to synthesize novel bacteriophages; EU AI Act took effect. For investors & developers, AI militarization & bio-design accelerate, yet regulatory compliance costs will rise in tandem.
Source ↗ - Other
🤖 Report: Demis Hassabis Also Wants to Leave Google DeepMind
Exclusive: Demis Hassabis, despite serving as DeepMind chair and Alphabet chief scientist, still eyes leaving Google DeepMind. For Alphabet, this marks another potential core executive departure after its chief scientist, straining its AI talent retention strategy.
Source ↗ - Others
🤖 Gartner: Half of China's AI accelerators to be domestic by 2030
Gartner predicts half of China's AI accelerators will be domestically produced by 2030. This signals a major acceleration in local compute independence; developers should watch open-source LLMs like Kimi K3, GLM-5, and DeepSeek V4.
Source ↗
+7 more headlines
- 🤖 Kimi K3 IQ2-XXS Model Slims from 711GB to 478GB, Drops Multilingual Support
- 🤖 Qwen 35B A3B vs Gemma 26B A4B Tokenization: Why Qwen Handles HTML/JS Better?
- 🤖 Firebird's Global AI Factory Push: 2GW Reserved, GPU Density +40%
- 🤖 Alphabet reshuffles Hassabis: DeepMind shifts from "frontline" to "strategy"
- 🤖 DeepMind CEO Hassabis may follow Jeff Dean in exit
- 🤖 MiniMax H3 Leads Video AI, DeepSeek Raises Prices and Nears $8B Valuation in Funding
- 🤖 DeepSeek's Price Hike Isn't Losses, It's Traffic Control