November update: A Safeguarded AI pivot + real-time climate monitoring
What’s new at ARIA
News and opportunities
Meet our incoming CEO, Kathleen Fisher. Beyond her tenure at DARPA, Kathleen has a track record of mobilising entire ecosystems – not just funding them. She’ll begin scaling our early momentum into global impact from February. Learn more.
Discover the latest from our Activation Partners:
Renaissance Philanthropy has announced their inaugural Ember cohort – turning big-if-true ideas into actionable R&D projects. Meet the cohort.
Venture Café is now live in Edinburgh. They welcomed ~260 people to their first event at the Edinburgh Futures Institute – find out more.
Early interest is open for the second cohort of Pillar VC’s Encode Fellowship, embedding top AI talent into UK labs to tackle challenges across our opportunity spaces. Sign up for updates – applications open 15 February.
Dig into our latest opportunity space, Collective Flourishing, where PD Nicole Wheeler sets out a vision rooted in our fundamental drive to seek better futures. Read it and share your thoughts.
Dive into our latest theses:
PD Rico Chandra’s programme thesis has evolved from Perpetual Flight into Enduring Atmospheric Platforms, with a tighter focus on long-endurance communication applications and an expanded range of technological approaches. Read it and share feedback.
In Massively Scalable Neurotechnologies, PD Jacques Carolan aims to develop a new class of surgery-free neurotechnologies, utilising the body’s natural pathways to reach the central nervous system. Read it and share feedback.
In Accelerated Adaptation, PD Yannick Wurm explores how advances in genomics, precision biology, robotics, and AI could unlock new pathways to enhancing ecosystem resilience. Read it and share feedback.
In Scaling Trust, PD Alex Obadia asks if AI agents could go out into the digital and physical worlds and mobilise, negotiate, and verify on our behalf. Read it and share feedback.
We’re hiring for three Technical Specialists across our Scalable Neural Interfaces and Trust Everything, Everywhere opportunity spaces. Learn more about being a T-Spec at ARIA and apply.
Opportunity seed funding is now open in Engineering Ecosystem Resilience, with rolling seed funding also available across Mathematics for Safe AI, Nature Computes Better, and Scoping Our Planet. Find all open funding calls.
Explore the opportunity seed projects we’re funding in Scalable Neural Interfaces – advancing the field by exploring novel biological materials, and pioneering new methods for interfacing with the nervous system. Dive into the projects.
Events
Discover upcoming events from our Activation Partners:
Cambridge NeuroWorks is bringing the NeuroTech Unconference to Manchester. Sign up to attend on 2 December.
Venture Café London’s next event will focus on collaborating for an inclusive AI future. Sign up to attend on 4 December and discover all December events.
Venture Café Manchester’s end-of-year event brings together innovation with workshops on communicating for your business and opportunities to connect. Join on 11 December.
Venture Café Edinburgh are hosting the regional round of Pitch2Tokyo, selecting the local innovator who’ll represent the city at the Venture Café Global Gathering 2026 in Tokyo. Get involved on 11 December.
Encode: AI for Science’s Winter Salon is taking place on 11 December. Register your interest.
AI progress and a Safeguarded AI pivot
While our conviction in the vision for the Safeguarded AI programme remains unchanged, the pace of frontier AI progress has fundamentally altered our path – instead of investing in specialised AI systems that can use our tools, it will be more catalytic to broaden the scope and power of the TA1 toolkit itself, making it a foundational component for the next generation of AI. We sat down with Programme Director David ‘davidad’ Dalrymple to unpack this pivot and learn what lies ahead.
What led to this pivot?
davidad: The decision was driven by the speed of progress in frontier AI capabilities. This has outpaced expectations from when the programme was designed.
If we look back over the last 12 months, every frontier model has more capability than I expected at that point in time. When METR, the nonprofit that measures capabilities, evaluated them, they quantitatively exceeded expectations in the same way. So, we’re not just updating based on one model; we’re updating based on a trend of the last four or five frontier models.
How did you decide that a pivot was needed?
davidad: Honestly, it didn’t occur to me that we could depart from the TA2 Phase 2 selection process so close to applications closing. But in a conversation with Nora, the Technical Specialist on the programme, and Ant, ARIA’s CTO, I mentioned that it was more likely than not we’d select zero successful teams for Phase 2 – the value of having a dedicated team pursuing ML R&D to develop the capabilities needed for the Safeguarded AI workflow seemed much lower than when we started, because we now expect many of these capabilities to come online soon by default.
Ant said, “What if we just stop it right now?” Once I realised that was a real option, I quickly agreed. We then had a call with Ilan, ARIA’s CEO, and he said, “Great, let’s do it.”
Did this experience of programme management surprise you?
davidad: One of the things I appreciate about ARIA is that they back Programme Directors to steer into new directions before the picture is complete. I’m grateful we were able to move so quickly on aborting the TA2 selection, while also having space to develop the new vision thoughtfully.
As part of the pivot, you indicated wanting to expand the scope and ambition of TA1…
davidad: Yes, instead of narrowly targeting assurance for cyber-physical system control, we are thinking about TA1 as a toolkit for mathematical assurance and auditability across a wider range of areas, including software and hardware verification, auditable multi-agent systems, and more informal knowledge.
Can you give us an example of one of those challenges?
davidad: My current leading hypothesis for a proof point for the platform we’re building is formally-verified firewall. Specifically, securing point-to-point connections over the open internet.
In critical infrastructure – power grids, water systems, radar and air traffic control – connections that once used dedicated wires now use the internet, which opens up cyber vulnerabilities. With the rise of general AI, more actors are capable of sophisticated cyberattacks.
So, part of protecting the public from AI risks is hardening critical cyber systems. DARPA proved a decade ago (through the HACMS programme led by Kathleen Fisher, our incoming CEO!) that formally verified software can be made meaningfully secure. With AI, we may be able to scale that process – faster and cheaper.
Why do these feel like the most useful application areas for these tools?
davidad: There are two different reasons. First, with shorter timelines to societal change from AI, we need to make an impact sooner. Feedback loops in information technology are much faster. Second, as AI systems become capable of autonomously executing increasingly complex and lengthy R&D tasks, we want to build a toolkit that enables high quality human oversight over those processes, while also being itself extensible by AI systems themselves.
How do you see the toolbox and frontier AI development interacting?
davidad: Frontier AI companies are eager for new, challenging tasks to train the next generation of models. A suite of tools and examples – like verifying a firewall or crypto protocol – could be invaluable for them. We’ll release the toolkit open source, so frontier AI labs can use it to advance formal reasoning capabilities.
This isn’t just, ‘can you write Python code that passes tests?’, but ‘can you write code that carries a proof verified by a trusted kernel?’ That trusted kernel also gives feedback that’s much harder for an AI system to ‘reward hack,’ since it must prove correctness for all cases – not just test cases.
It’s a win-win: if the next generation of AI models becomes more skilled at proving formal properties using our tools, those same tools will be even more valuable when those models are released.
Find out more about Safeguarded AI.
Activation Partners: Seeding new AgTech ventures with Nucleate UK
Our Activation Partners play a vital role in helping the full potential of our opportunity spaces to be realised – helping to create the conditions for bold ideas to thrive, building and connecting communities, and empowering those driving impact beyond the lab.
One of these partners is Nucleate UK: a student-led, non-profit organisation empowering a global community of bio-innovators. They promote science commercialisation to researchers by building more consistent and smoother spin-out pathways, early entrepreneurship education, and increased financial and operational connectivity across the ecosystem.
A core aim of ARIA and Nucleate’s partnership is supporting our Programmable Plants opportunity space through Nucleate’s Eco track. The track fosters biotech entrepreneurship aimed at tackling sustainability and climate challenges, offering masterclasses, workshops, and its flagship Activator programme to help early-stage science transition into commercial ventures.
The first ARIA-powered UK Eco cohort ran from October 2024 until March 2025 – ARIA’s support empowered Nucleate UK to have the biggest cohort amongst the 22 chapters running an Activator programme. Kicking off with a series of events across the entire country to build a community from scratch, our partnership is already catalysing scientific and economic impact in the UK. Success stories from the first cohort include Bindbridge, whose founders joined with a concept of computationally accelerating molecular discovery for agriculture. Over the course of the programme they went from idea to incorporated company which has now raised a multimillion pre-seed round in investment from investors including Nucleus Capital and Speedinvest.
Another is 2Blades, an AgTech organisation from The Sainsbury Laboratory in Norwich focused on delivering durable disease resistance in crops. 2Blades joined Eco to explore the best strategic path for translating their fundamental R&D into real-world applications. They ultimately concluded that remaining a non-profit was the best way to preserve their mission and impact. ARIA and Nucleate see this as a success: “A core purpose of the Activator is to give teams a structured, time-bound way to pressure-test startup ideas and company models as quickly and honestly as possible,” says Federico Caso, who directs the partnership on behalf of Nucleate UK. “Sometimes that results in a new, venture-backed company like Bindbridge; sometimes it results in a well-informed decision to continue in other directions.”
Building on the success of the first Activator, the partnership is growing the Eco track community through technical hackathons, UK-wide roadshows and foundational education workshops in order to catalyse new ideas, talent and capital into the Programmable Plants opportunity space. “By supporting Nucleate UK’s Eco track, we’re not just helping to build our own Programmable Plants space, but we’re actively catalysing a new, dedicated pillar of entrepreneurial activity in the ecosystem,” says Pranay Shah, Product Manager at ARIA and lead for our Activation Partners initiative. “Biotech translation for climate impact currently lacks the robust entrepreneurial pathway seen in human health areas such as drug discovery – our vision is to rapidly expand this through our partnership with Nucleate, ensuring a sustained flow of exceptional founders, breakthroughs, and investment that will endure long after ARIA’s initial work in the space.”
Looking forward, Nucleate is gearing up to further increase access to biotech entrepreneurship around ARIA’s opportunity spaces. In 2026, our partnership will launch Catalyse, a fully online ‘Biotech 101’ programme designed to help researchers take their first steps toward company creation. Nucleate will also run dedicated Investor Roadshows, providing sustainability and climate focused biotech investors the chance to meet and interact with early stage Eco ventures across the UK.
When it comes to the goals of the partnership, “the clearest measure of success will be the impact we have on seeding a more ambitious, risk-tolerant entrepreneurial culture in the UK AgTech ecosystem,” says Fiachra Sweeney, Director of Eco at Nucleate UK. “One where researchers feel supported to pursue bold, unconventional ideas.”
Autonomous, scalable, real-time climate monitoring: A Q&A with Marble
Mathieu Johnsson and his team at Bristol-based startup, Marble, are leading the Arctic DronePort project – part of our Forecasting Tipping Points programme.
The team is looking to enable real-world environmental monitoring through a distributed network of unmanned hangars and fast, low-cost drones. As a first step, they’re monitoring the icefjord of Ilulissat – a coastal town in western Greenland – to better understand the accelerating retreat of the Sermeq Kujalleq Glacier and the sudden rolling of icebergs in Disko Bay.
We caught up with Mathieu at Greenland Science Week to hear how they’re innovating Arctic remote sensing and using their data for near-term climate forecasts.
Can you tell us more about the Arctic Droneport project?
Mat: We fly fleets of small high speed drones up and down the fjord and in the bay – they’re equipped with multiple sensors to reconstruct the ice sheet, glacier front and iceberg shapes in 3D.
We have a strong focus on rapidly prototyping the full chain from sensing to prediction to communication. Working with scientists from the University of Bergen and the Greenland Institute of Natural Resources, we turn that data into forecasts that we can then deliver to the people who are actually affected – from local fishermen to policymakers in Greenland.
What inspired you to work on this specific challenge?
Mat: First, ARIA itself. The way the organisation was describing how it wanted to work, and the type of people they were looking for – it felt exactly like me. Second, the Forecasting Tipping Points programme goes straight into something I care about personally: long term sustainability for both society and the planet. The Greenland ice sheet is one of the biggest global risks in that story.
Third, it fits perfectly with what we have always been building at Marble: advanced drone fleets that can cut the cost of collecting high quality data about the surface of the Earth by orders of magnitude. We want to make this data accessible. ARIA lets us push that mission directly into one of the most important climate problems we face.
What do you wish more people knew about your research area?
Mat: I wish more people really understood the scale of the risk we are dealing with and how close it is to our lifetimes. Changes in the Greenland Ice Sheet and the North Atlantic currents could realistically mean that a country like the UK – which sits at the same latitude as Canada – ends up with a Canadian climate. Most people in Europe have no idea how disruptive that would be.
I also wish people realised how poor our actual visibility of the planet is compared with what films and Google Maps make us think. We do not have real-time, high-resolution coverage. Most satellites pass every few days, at low resolution and only if the weather is good – that gap is part of the reason why we still do not fully understand how the Greenland Ice Sheet is melting. Robotics lets us move sensors around at almost zero marginal cost, and that is a huge, still underused opportunity.
What have you learned from Greenland Science Week?
Mat: Being here has shown me how many other valuable use cases could sit on top of the system we’re building if we deploy it permanently: from whale and narwhal monitoring, to new ideas for stitching drone data into better 3D models, and even ways to use drones to lower instruments into the water for temperature and salinity measurements.
The big positive has been the serendipity of being in the same place as so many people who think differently about similar problems. On the flip side, I’ve also seen how much of the field still runs on short-term, project-by-project funding and a ‘science first, impact later’ mindset. To really tackle something as big as Greenland’s melt, we need sustainable business models and broader use cases alongside pure science, which is exactly the culture ARIA is trying to push toward.
How have you approached engaging with the Greenlandic public and why is it so important to your ARIA work?
Mat: Engaging with people in Greenland has been central from the beginning.
There’s a basic responsibility – we need to build relationships and break the pattern of international scientists arriving, measuring, then leaving. Importantly, the people in Greenland are the ones who will feel the impact first. That’s why we’re focusing so much on the iceberg rolling problem in Ilulissat: these events can generate mini-tsunamis that endanger fishers and people onshore. This focus lets us test the whole loop: from collecting data, to building models and designing interfaces that local communities might actually use and trust when they decide whether to go out on the water.
Looking ahead, we want our systems to be permanently deployed in Greenland and eventually beyond, and supported by a mix of scientific and commercial use cases, like resource monitoring. We want to build long term partnerships in Greenland now and inspire the young people who will one day operate, maintain and even redesign these systems for their own needs.
Discover all the projects funded in Forecasting Tipping Points.
F-Spec Corner: Recommended reads
ARIA’s inaugural cohort of Frontier Specialists (F-Specs) are a small, dedicated team with the mission to dramatically expand ARIA’s technical surface area and sharpen the cutting edge of the science we’re funding. Here are some of the pieces that F-Specs Edith-Clare Hall and Matt Burnett have been digging into this month:
Biohybrid Actuators in Robotics
Even as our growing set of autonomous machines are deployed to perform tasks in the real, messy, unstructured, biologically integrated world, we constrain their forms to rigid materials and brittle, centralised code. But what if the robot’s very power source offered the solution? What if we delegated the messy, continuous work of motion and adaptation not to an algorithm, but to living muscle and, hence, leveraging biological adaptability to generate transient, task-specific forms of intelligence? This is a high quality in depth review of biohybrids that details the opportunities of integrating skeletal and cardiac muscle tissues with synthetic materials, thus shifting from programmed control to embodiment-centric performance.
Multifunctional Fluidic Units for Emergent, Responsive Robotic Behaviours
Another response to the challenge of rigid robotics in the dynamic real world is Mousa et al.’s recent work on creating the mechanical equivalent of an electronic logic gate. This modular fluidic unit enables the delegation of complex sensing, switching, and actuation duties to the physical structure itself, bypassing traditional control systems by distilling all functionalities into one reconfigurable, hysteretic unit. When configured as a responsive self-oscillating actuator, the unit combines all three roles at once, with its oscillation frequency directly determined by the applied counter-moment (external load), effectively making it a self-sensing oscillator. Most excitingly, when multiple units are mechanically coupled via a shared body, their individual oscillations passively synchronise through implicit physical interaction. This capability allows the robot to self-organise into coherent locomotion (like a forward galloping gait) and execute decision-making (like a crawler that stops at a boundary) entirely via fluidic logic encoded within the mechanical structure, offering an alternative to conventional electronics.
Seb Krier sets out a vision where externalities are efficiently priced into decision making, made possible by AGI. In this vision of the future inflexible, prohibitive regulation is replaced by instantaneous bargaining by representative AIs in which everyone is a winner. This detailed op-ed unpacks both the why and the how AGI might unlock new forms of governance, business and collective flourishing. It also dovetails nicely with one of ARIA’s new programme theses, ‘Scaling Trust’.