A mission control for frontier technology
UK Horizons: FRC — Renaissance Philanthropy – A brighter future for all through science, technology, and innovation
Successful coordinated research programmes, like those at ARPAs, have often resulted from exceptional contractors. Many of the best contractors leveraged organisational models uncommon in academia or VC-funded startups. They combined technical ambition, a willingness to build real technology for actual users, and flexible management structures all under one roof to tackle ambitious R&D problems which made the programs a success. These organisations were driven by the pursuit of an ambitious North Star technical vision, not revenue-maximisation. And they pursued this vision with a mix of contracts and grants. We call organisations that do this Frontier Research Contractors (FRCs). Two of ARPA’s most impactful programmes, the original internet and autonomous vehicle projects, were both led by FRCs. FRCs are also often referred to as 'BBNs' by researchers, inspired by the early internet (ARPAnet) contractor, Bolt, Beranek & Newman.
Poolside: Frontier research to operational intelligence
Poolside is a foundation model company bringing intelligence to everywhere work gets done. Our mission is to drive abundance for humanity by creating artificial general intelligence.
pi-ds4 · Audrey Tang
Run a frontier model on your own machine with stable, contestable decision traces. Full install, steering, reproducibility, and tuning guide.
NASA
Intro to NSF Tech Labs: Session I
Pluralis Research
Protocol Learning — decentralized, communication-efficient model-parallel training for foundation models.
Intro to NSF Tech Labs: Session II
Space Exploration Technologies - S-1
Pacing the Frontier
A statement from over 1000 employees of frontier AI companies
Cheating behaviour in frontier model evaluations | AISI Work
We find cheating behaviour in all of our cyber capability evaluations, and outline the implications as models grow more capable.
.png)
Building an Advanced Agentic Harness | Data For Science
From a single pilot to an air campaign: planning, parallelism, memory, verification, and observability for production-shaped agents.
A snapshot of research into answering if frontier AI agents can run R&D into AI (which not surprisingly failed apart from "minor findings" and "engineering steps"). The paper lists its TBDs strengths and limits, worth a read before people who don't know theirs jump in arxiv.org/pdf/2607.27191
