METR, or Model Evaluation and Threat Research, is a nonprofit research institute located in Berkeley, California. Founded in August 2022, METR focuses on evaluating advanced AI models to identify capabilities that may pose significant risks to society. The organization conducts pre-deployment empirical evaluations of AI systems, assessing dangerous capabilities such as autonomous replication and cybersecurity threats. METR's mission is to develop scientific methods for assessing the risks associated with AI systems' autonomous capabilities. It provides services to leading AI companies, including OpenAI, Anthropic, and Google DeepMind, helping them understand AI capabilities and risks before deploying new models. The organization also contributes to the development of standardized evaluation methodologies and publishes research to enhance public understanding of AI risks. With a dedicated team, METR aims to promote safe AI development and informed decision-making.
METR, or Model Evaluation and Threat Research, is a nonprofit research institute located in Berkeley, California. Founded in August 2022, METR focuses on evaluating advanced AI models to identify capabilities that may pose significant risks to society. The organization conducts pre-deployment empirical evaluations of AI systems, assessing dangerous capabilities such as autonomous replication and cybersecurity threats. METR's mission is to develop scientific methods for assessing the risks associated with AI systems' autonomous capabilities. It provides services to leading AI companies, including OpenAI, Anthropic, and Google DeepMind, helping them understand AI capabilities and risks before deploying new models. The organization also contributes to the development of standardized evaluation methodologies and publishes research to enhance public understanding of AI risks. With a dedicated team, METR aims to promote safe AI development and informed decision-making.
Interested in this position?
Apply directly on the company website
The AI Strategy & Transformation Architect at NeuraFlash, part of Accenture, guides enterprise clients from AI strategy through Salesforce-based implementation by converting ambiguous business challenges into actionable roadmaps and working software.
SpaceXAI is seeking a multilingual AI Tutor specializing in Italian audio to curate and annotate speech data that improves Grok’s voice interactions, speech recognition, and auditory experiences across languages and accents.
Mindrift, powered by Toloka, is seeking experienced consultants from top-tier strategy firms to design realistic consulting environments, AI tasks, and evaluation frameworks that teach advanced models business reasoning.
SpaceXAI is hiring an AI Tutor focused on multilingual audio to help train Grok for better voice interactions, speech recognition, and audio understanding across languages and accents.
