Anthropic Launches Initiative to Fund Advanced AI Benchmark Development

The program aims to address the limitations of current evaluations and enhance AI safety and capability assessments.

Anthropic invites third-party proposals to create robust benchmarks for evaluating AI models.
The initiative will focus on AI safety levels, advanced capabilities, and societal impacts.
Funding will support the development of tools, infrastructure, and large-scale evaluation tasks.
The program seeks to measure AI's ability to handle complex tasks and resist malicious use.
Anthropic's effort highlights the growing need for comprehensive AI safety and performance assessments.