🤖 AI HOT Daily · Jun 11, 2026
Google DeepMind releases DiffusionGemma 4x faster text generation, Xiaomi launches MiMo Code V0.1 open-source coding assistant, EU orders Meta to open WhatsApp to third-party AI, Google held liable for AI hallucinations.
1. Google DeepMind Releases DiffusionGemma
Open-source diffusion model breaks autoregressive token-by-token generation, producing 256 tokens per forward pass. 26B MoE activates only 3.8B params, fits 18GB VRAM after quantization, 4x faster.
2. Xiaomi Releases MiMo Code V0.1
Xiaomi's open-source terminal AI coding assistant supports million-token context, infinite context, agent-model deep collaboration, Compose mode, voice input, zero-cost migration from Claude Code.
3. Grok Voice Delivers Human-Like Speech at Low Cost
xAI launches Grok Voice with human-like timing, tone, and warmth at a fraction of competitor pricing.
4. Cursor Bugbot Update: 3x Faster
Cursor's Bugbot code review tool runs 3x faster with 22% lower cost, finds 10% more bugs per review, 90% complete within 3 minutes.
5. Huawei Cloud Launches CloudRobo Embodied AI Platform
World's first end-to-end embodied AI development platform covering the full lifecycle from data to deployment, built on PB-scale trusted data.
6. EU Orders Meta to Open WhatsApp to Third-Party AI
EU Commission orders Meta to give third-party AI assistants free access to WhatsApp during antitrust investigation to prevent market competition harm.
7. Anthropic: AI Builds Exploits in Hours
Anthropic finds its AI can turn security patches into working exploits in hours at minimal cost without expert knowledge, declaring traditional patch cadence obsolete.
8. Google Held Liable for AI Hallucinations
A court ruling holds Google legally responsible for AI model hallucinated content, potentially creating major precedent for the AI industry.
9. China Ministry Issues AI+ICT Development Guidelines
Ministry calls for accelerating 400G/800G backbone networks, promoting 5G-A/6G and optical network integration with AI, encouraging AI-empowered telecom services.
10. Baidu & Fudan's LU-KV Framework Accepted at ICML 2026
Long-horizon Utility KV framework models KV cache allocation as global combinatorial optimization, achieving strong results on LongBench and RULER benchmarks.
Sources and verification
This legacy briefing did not retain its original source URLs. Verify safety, financial, policy, and breaking-news claims with authoritative primary sources before acting on them.