Scale AI announces the acquisition of ICG Solutions to accelerate the U.S. Government's AI-First transition for national security. By combining ICG’s LUX™ platform with Scale’s agentic platform, we are delivering an unmatched, end-to-end intelligence solution for the Department of War and the Intelligence Community.

Read more

Can AI turn a novice into a biosecurity expert? Our study with SecureBio found that non-experts using LLMs outperformed trained biologists, and current safeguards barely slowed them down.

Read more

Enterprise data is scattered across systems, formats, and contexts that were never built for AI. Agents are forced to reconstruct the organization from fragments, leading to brittle workflows and unreliable results. The AI-native data layer, the first pillar of Dialect, unifies these sources into a connected semantic foundation that agents can consistently understand and reason over in production.

Read more

Health chatbots are one of pharma's most common first steps into GenAI, but a red team evaluation of a live deployment revealed safety gaps that standard QA processes aren't built to catch.

Read more

Your business runs on context, exceptions, and institutional judgment—elements that are opaque to standard LLMs. Introducing Dialect: an infrastructure layer that bridges the gap between fragmented data and autonomous agents. Learn how the "Enterprise Intelligence Loop" turns expert feedback into a compounding competitive advantage.

Read more

A new collaboration between Scale AI and BAE Systems Intelligence & Security sector is designed to accelerate the development and fielding of advanced artificial intelligence capabilities into the architecture of the U.S. Department of War’s most capable current and future military platforms and systems.

Read more

EgoVerse introduces a scalable approach to training Physical AI by combining high-fidelity human demonstrations with robot data in a shared learning framework. Built with leading academic and industry labs, it releases both a large egocentric dataset and an open-source pipeline, showing that co-training across human and robot data improves performance, generalization, and transfer across different robot systems.

Read more

Voice Showdown is a preference-based benchmark for voice AI models, built on real human speech to measure how models perform across languages and real-world conversations.

Read more

Read more