Algorithmic Research Group builds tools and infrastructure for AI safety research. Benchmarks for evaluating autonomous agents. Datasets for studying how models fail. Runtimes for running agent workloads at scale.
Small team, open output
The lab is run by Matthew Kenney and a rotating cast of contractors. One person works full-time, with specialists brought in as projects demand.
We build benchmarks, datasets, and infrastructure for evaluating what autonomous AI systems can actually do. The research output is open source so other researchers, auditors, and teams can inspect and use it.
For-profit, public-interest
Algorithmic Research Group is a for-profit research lab. Consulting work and sponsorships fund its open-source research. The structure keeps the work lean and self-directed; publishing the output makes it available for independent evaluation.
Support the research
Sponsorships fund open-source benchmarks, datasets, and tooling for AI security research. Everything built with sponsor support is released publicly.
Sponsorships are not tax-deductible.
Work together
For funded research projects, dataset sponsorships, or infrastructure partnerships, start a conversation. We're flexible on structure and committed to keeping the research output open.