Topic
Frontier models
Frontier models.
Frontier models are general-purpose AI models developed at the current edge of the field. Our stories report that New York's RAISE Act, signed in December 2025, requires AI developers to publish and follow safety plans and report incidents to the state, as covered on November 16, 2026. A shared story from November 6, 2026 describes distillation attacks, which copy an AI model's skills by harvesting its answers at scale, and Anthropic's defenses dated October 12, 2026. Our coverage also includes State of AI Report 2026, published October 8, which reports AI doing 26% of Claude's R&D work and inference cost 47% lower in a quarter.