フランスのAI企業Mistral AIが2026年8月4日、文章や画像が指定した安全基準に当てはまるかを判定するAIモデル「Shieldstral 1.0 3B」を公開しました。パラメーター数は30億で、サービスごとの安全基準を自然な文章で指定できる点が特徴です。Introducing Shieldstral. | Mistral AIhttps://mistral.ai/news/shieldstral/Introducing Shieldstral, Mistral’s 3B open-weights m…
[株式会社アトラックラボ] [画像: https://prcdn.freetls.fastly.net/release_image/52796/117/52796-117-bd8b32bd102103de459c6f3c2301a9d0-2025x2700.jpg?width=536&quality=85%2C75&format=jpeg&auto=webp&fit=bounds&a...
Guardrails for AI agents, with a receipt. Blocks destructive tool calls (rm -rf, terraform destroy, force push, DROP TABLE, leaked keys) before they run, and signs every decision into an off-box, hash-chained record anyone can verify.
A fast algorithmic trading platform with support for AI based strategies
Glencore Plc reported a steep jump in profits after a surge in the price of its most important commodities and its trading unit had one of its best ever periods as the Iran war sent energy prices soaring.
