0
Applied AI·July 2, 2026·1 min read

US in talks with AI companies over voluntary standards for new models

Share

Voluntary US standards on model release — benchmarks, timelines, and disclosure rules — would turn today’s ad hoc safety practices into a soft compliance baseline for anyone shipping advanced models. If you're building on third-party LLMs, expect more structured release notes and evals, but also potential lags between capability readiness and public availability.

Applied AI

Sources: Anthropic has initiated early-stage development of a custom AI server chip and held preliminary discussions with Samsung about manufacturing the chip

Another frontier lab moving into custom AI silicon means model providers are treating chips as a strategic lever, not a commodity input. For large buyers, this could translate into tighter integration between models and hardware — and more vendor lock-in — so negotiate for portability and multi-cloud options early.