A positive safety case requires the technical tools to support it. They do not yet, at least not at the reliability that high-stakes deployment demands. Closing that gap is the most important contribution that research funders can make to AI monitorability. We propose an independent monitorability research fund, modelled on the UK’s AI Alignment Project. It would be co-financed by philanthropic organisations, frontier AI providers, and EU public funding bodies like Horizon Europe. The fund should support three priorities: strengthening the benefits of existing CoT monitoring approaches, since they remain the most effective oversight tool today; proactive research on monitoring methods for opaque architectures – internal probes, learned feature detectors, causal interpretability techniques – conducted before such architectures become dominant rather than after; and scaling work, extending interpretability methods that succeed in the lab so that they can work at the scale, and in the adversarial conditions, of frontier deployment. Funders could also set up pull mechanisms such as milestone prizes, competitions where researchers develop monitoring methods and validate them on standardised testing infrastructure, and bounties for reproducible failures of proposed monitoring techniques. These can complement the fund by drawing academic, independent, and open-source researchers into research areas that are not profitable enough for frontier AI companies to prioritise.