creators

Beth Barnes - Founder and CEO of METR

Beth Barnes founded METR, the Berkeley nonprofit that tests frontier AI models for dangerous capabilities and publishes the AI task-horizon trend.

Beth Barnes previously worked with DeepMind's chief scientist on scaling laws for forecasting deep learning progress, and later at OpenAI, where she helped develop safety targets, evaluated scalable oversight techniques for alignment, and assessed models for misalignment before release. She left OpenAI in 2022 to found what became ARC Evals, the evaluations division of Paul Christiano's Alignment Research Center. In December 2023 ARC Evals was spun out as an independent nonprofit and renamed METR (Model Evaluation and Threat Research), based in Berkeley, California, where Barnes is founder and chief executive. METR runs pre-deployment evaluations of frontier models for dangerous capabilities, including deception, self-replication and evading shutdown. Its most-cited result measures the length of task an AI agent can finish on its own; a January 2026 revision put the doubling time at 196 days across the whole series and 131 days since 2023. METR also published a randomised trial finding experienced open-source developers took 19% longer to complete issues when using early-2025 AI tools. Barnes worked with Anthropic on its Responsible Scaling Policy, and was named to TIME's 100 most influential people in AI in 2024 and again in 2026.

Specialization

AI evaluations, dangerous capability testing, AI safety, autonomy benchmarks

Country

United States