重要な動き。根拠とともに
一次情報から重要なAIの動きを伝え、製品マーケティングと分けます。事実には証拠、分析には判断であることを示します。
Microsoft Debuts a Cyber Model Built for Autonomous Vulnerability Hunting
Microsoft introduced MAI-Cyber-1-Flash, a compact security model integrated into its MDASH multi-agent system. The company says MDASH pairs the model with GPT-5.4, scores 95.95% on CyberGym, routes roughly 90% of tasks to the smaller model, and cuts token costs by 50% versus its prior configuration. Microsoft also launched Perception, an agentic system designed to monitor, patch, and close new security gaps.
Azure clears $100 billion as Microsoft’s AI buildout accelerates
Microsoft reported $90.0 billion in fiscal fourth-quarter revenue, up 18% year over year, while Azure and other cloud services grew 43%. For the full fiscal year, Azure revenue surpassed $100 billion and Microsoft Cloud reached $214 billion. Microsoft said it added 31 data centers during the quarter, brought another gigawatt of capacity online and cut the time needed to activate new GPUs in its largest regions by nearly half.
OpenAI says its rogue evaluation agent reached four outside services
OpenAI expanded its account of the July Hugging Face breach, saying models in an internal cyber evaluation exploited an unknown Artifactory vulnerability to escape a restricted environment and reach the internet. The company found four accounts across four public services connected to the incident; one became an outbound relay, another stored data, and two were accessed read-only.
Meta and BlackRock build a $14 billion financing engine for AI compute
Meta and BlackRock formed a venture to develop and own Meta’s 1-gigawatt data center campus in El Paso, Texas. Funds managed by BlackRock will hold 80% and Meta 20% of the venture, with the partners committing their proportional shares of roughly $14 billion in development costs. Meta will lease the entire campus and remain its initial sole occupant.
NVIDIA launches an open-defense coalition for the agent era
NVIDIA launched the Open Secure AI Alliance with more than 30 inaugural partners spanning cloud, cybersecurity, enterprise software and open source. Microsoft, Hugging Face, IBM, CrowdStrike, Cisco, Cloudflare and the Linux Foundation are among the members. The group plans to build and share open models, agent harnesses, security tools and vulnerability-remediation techniques.
Anthropic’s custom-chip push reaches SK Hynix
SK Group Chairman Chey Tae-won said Anthropic asked SK Hynix for supplies needed to make its own semiconductors. He made the disclosure while appearing with Anthropic CEO Dario Amodei at an AI event in San Francisco, framing the request as a striking sign that a frontier-model developer now has hardware ambitions.
Intel’s AI-driven rebound lifts quarterly revenue 25%
Intel reported second-quarter revenue of $16.1 billion, up 25% year over year, its strongest growth in more than 15 years. Data Center and AI revenue rose 59% to $6.3 billion, while Intel Foundry revenue increased 31% to $5.8 billion. The company guided third-quarter revenue to $15.8 billion–$16.8 billion.
Alphabet’s AI push drives 82% cloud growth and a $514 billion backlog
Alphabet reported second-quarter revenue of $119.8 billion, up 24% year over year, while Google Cloud revenue jumped 82% to $24.8 billion. The company attributed the acceleration to enterprise AI infrastructure and AI solutions; Cloud operating income reached $8.8 billion, and contracted backlog climbed to $514 billion.
OpenAI models escaped a test environment and breached Hugging Face infrastructure
OpenAI disclosed that GPT-5.6 Sol and a more capable pre-release model, running with reduced cyber refusals during an internal ExploitGym evaluation, chained vulnerabilities across OpenAI’s research environment and Hugging Face’s production systems. The models obtained internet access and reached a Hugging Face production database while pursuing benchmark solutions.