AIニュース Nov 4, 2026

今日の注目点

• AI policy and midtermsVoters backing data center moratoriums in the 2026 midterms met a spending push from AI industry PACs, while NBC News polling found 76% called AI and data centers important even as exit polls ranked AI last among campaign issues. Infrastructure politics and lobbying now move together.
• AI chips and computeDeepSeek reportedly plans to train its next V5 model on Huawei Ascend chips, extending Ascend from inference to training, while AMD agreed to pay about $2.9 billion for Toronto startup Taalas. Chip supply and silicon ownership remain central to model competition.
• Model control and verificationMIT researchers reported GPT-6 Astra could rewrite another model's weights, lifting a Qwen model's math accuracy from 7.9% to 48.1% by changing 0.016% of weights, while Anthropic's FlashProve prototype uses cryptography to verify which model produced a response. Model control and provenance are drawing research attention.
• Capability releasesOpenAI says an autonomous pipeline found, validated and patched 948 Linux kernel privilege-escalation vulnerabilities, while Tencent released Hy4 with vision and computer-use features and Ricursive published Ripple-65B for agentic coding. Capability gains are arriving across security, multimodality and open weights.
• Funding and M&AFunding concentrated on infrastructure and data control: fal raised $1.3 billion at a $16 billion valuation with NVIDIA participating, and Sentra raised $150 million for monitoring AI agent data access. Capital is following compute and governance layers rather than applications alone.

AI 生成分析