February update: Accelerating adaptation and enabling trust between LLMs
What’s new at ARIA
We’ve committed £50m to create the Scaling Inference Lab – an AI testbed within our Scaling Compute programme designed to prioritise rapid iteration, open collaboration, and long-term sustainability. Find out more and express your interest for future participation and read more about the Scaling Inference Lab in Post Haste.
Discover funding across our new programmes:
Accelerated Adaptation is exploring pathways to accelerate the adaptation of wild species to prevent biodiversity loss, and secure the natural infrastructure underpinning our global economy and wellbeing. Discover the programme and submit a concept paper by 6 March.
Universal Fabricators seeks to harness proteins to produce a functionally universal range of materials at scale. Find out more and submit a concept paper by 9 March.
Massively Scalable Neurotechnologies aims to develop a new class of neurotechnologies that reach and interact with the central nervous system via the body’s natural pathways. Read the call and submit a concept paper by 17 March.
Scaling Trust is focused on enabling AI agents to securely coordinate, negotiate, and verify with one another on our behalf. Learn more and apply for funding by 24 March.
Full proposals are also now open for Enduring Atmospheric Platforms. Help us to unlock a digital infrastructure layer between the Earth and space – apply for funding by 2 April.
We’ve published a concept note for our next Activation Partner (AP) cohort. Dive in and share your thoughts and explore opportunities across our APs below:
Applications are open for the third cohort of Fifty Years’ 5050 programme – a 14-week programme designed to help aspiring founders launch deep tech startups, running March-June. Apply now (rolling).
Become an Encode: AI for Science Fellow and spend a year with the freedom to build what matters. Starting in September, fellows will get the support to turn their projects into a leading startup, FRO, or open-source initiative. Apply by 28 March.
Connect with startups, researchers, investors and more at Venture Café’s Thursday Gathering events across London, Manchester and Edinburgh.
Venture Café London: Scale Smarter (5 March), Creative Catalyst (12 March), Launching Bind Research (19 March), Innovating Health (26 March)
Venture Café Edinburgh: Scotland’s Next Economy (12 March)
Venture Café Manchester: AI Social Club (5 March), AI + Automation (19 March)
Join Cambridge Neuroworks for Brain Mapping + Neural Connections – an exploration of the brain with expert talks, panel discussions, networking and insights into emerging technology pathways. Sign up for the event on 18 March.
Expanding the Conservation Toolkit
Earlier this month we launched the Accelerated Adaptation programme, backed by £54m. With concept papers now open, Programme Director Yannick Wurm shares why we’re exploring pathways to accelerate the adaptation of wild species.
For much of my scientific life, I was instinctively wary of deliberate human interventions in nature. Natural systems are highly complex, and I (like many others) have been frustrated by the hubris of those who have underestimated the risk of things going wrong. Countless ‘no-risk’ interventions have had detrimental consequences; even well-intentioned conservation breeding programmes can increase population numbers while collapsing genetic diversity and reducing long-term resilience.
I’ve spent most of my career as an academic, mostly focused on understanding the tradeoffs and genetic bases of social evolution in ants – the world’s coolest 20,000 species (that’s a discussion for another day). But having initially trained as an engineer I had a latent desire for more immediate impact. Driven by my frustration that we have struggled to reverse pollinator declines, I realised that new technologies could help fill data gaps and, in turn, address knowledge gaps. My team at Queen Mary University of London also pioneered the development of high-resolution approaches for measuring pollinator health.
Joining ARIA last year as a Programme Director provided a new opportunity for impactful work. How would you steward £50m towards ‘edge of the possible’ research that could change the world? My initial aim was adjacent to my academic research: could we reverse-engineer the secrets of the ants, wasps, and termites to reveal the molecular mechanisms underpinning their impressive feats? How can a queen live orders of magnitude longer than her genetically identical siblings? How are social insects resilient to pathogens despite living in the dirt and eating raw prey? I considered expanding to other organisms – from the ability of extremophile yeti crabs to thrive on high-pressure high-temperature deep-sea vents, to the planarians and salamanders that can regrow body parts or entire bodies. Or finding ways of making agriculture more sustainable. Or predicting evolution, which could impact applications from cancer treatment to pharmaceutical production to wastewater processing.
ARIA’s process kept forcing me to think about humanity’s greatest challenges, and that brought me back to the relationship we have with nature itself. Through hundreds of conversations, >15 workshops, and by funding ARIA’s first 22 discovery projects, I was able to further appreciate how critically vulnerable our ecosystems are, how urgently humanity needs to stop and reverse the unprecedented pressures we are creating, and how vital it is to expand our existing toolkit to support nature. My engineering instinct kept surfacing: could we become as intentional in our relationship with nature as we are when we build buildings our algorithms?
There are so many challenges in this space (see our Engineering Ecosystem Resilience opportunity space). ARIA-scale programmes could focus on better measuring living organisms, or better data modelling and integration. But simply getting better at quantifying declines is insufficient. As we get better at pinpointing ecological vulnerabilities, and risks of collapse, we need to face the reality that new types of interventions may become an essential part of the toolkit to halt further irreversible decline.
This mindset shift wasn’t easy – 18 months ago I would never have thought that I might be pushing for the ideas outlined in the Accelerated Adaptation programme.
At the heart of this change is my belief that the balance of risks has shifted. Over recent decades, risks of biodiversity loss and ecological collapse have dramatically increased – now one in four animals and plants face extinction within a century. These are not fringe concerns; the World Economic Forum cites biodiversity loss and ecosystem collapse as the second largest threat that humanity faces over the next 10 years. And UK GDP is already expected to shrink up to 3% due to chronic nature degradation – that’s before considering the potential impacts of reaching any ecological or climatic tipping points.
The risks of intervention remain real, but the risks of maintaining the status quo have grown larger still. The question is no longer whether intervention is risky, but whether continuing on our current path guarantees failure. Building a new set of tools that can complement and augment well-established conservation approaches could provide the step-change required to help transform the impact of our conservation efforts.
We can already see the direction of consensus beginning to shift. Most notably, the International Union for the Conservation of Nature (IUCN) decisively adopted a landmark policy framework for the responsible, science-based use of synthetic biology in nature conservation (88% in favour of Motion 87). And work on coral reefs provides a key example – in the face of total collapse as a result of rapid warming and acidification, conservationists who would once have rejected the most extreme forms of active intervention are now reconsidering.
But what should breakthrough tools that could complement existing nature protection approaches look like? What we’re proposing with the £54m Accelerated Adaptation programme is not a single technology or an unquestioning embrace of biotechnology. It is a spectrum of approaches, many of which are still relatively new in conservation contexts. For example, could we place oak saplings in climate chambers simulating 2100 conditions and select the survivors for planting? Could we develop vaccines that protect amphibians against the pathogens decimating their populations? Such approaches sit alongside transient physiological or molecular treatments that enhance stress tolerance without permanent genetic change. Crucially, the programme deliberately avoids sweeping, broad-brush approaches like uncontrollable gene drives aimed at collapsing populations. Our approach is not about domination of nature, or replacing well-established methods, but about expanding our toolkit in a disciplined, responsible way.
The boundaries of current regulatory frameworks, important ethical questions, and gaps in risk modelling all argue for caution. That is why the programme focuses on indoor, contained research environments, and why we are embedding ethical oversight from the outset — through an ethical advisory committee, dedicated ethics research teams, and a requirement that every funded project engages with the social and ethical dimensions of its work from day one. It is also why I have resisted narrowing the effort to a single species or a single philosophical stance. What matters most is outcome, not ideology. Different systems will require different approaches, and progress will come from constructive competition between the best ideas, regardless of where they originate. By programme end, we aim to have demonstrated measurable increases in resilience in at least two wild species in contained environments — and to have laid the scientific, ethical, and technical foundations for responsible real-world translation.
I did not abandon my scepticism. I refined it. The same instinct that made me wary of intervention now makes me wary of inaction; because the evidence no longer supports treating ecosystem collapse as something we can simply hope to avoid. The challenge before us is understanding how to intervene with humility, rigour, and an unwavering commitment to understanding risk.
If this resonates, we want to hear from you. We are looking for cross-disciplinary teams: ecologists, conservationists, evolutionary biologists, molecular biologists, engineers, ethicists, modellers — who believe they can make meaningful progress responsibly on accelerating adaptation in wild species.
Submit a concept paper for Accelerated Adaptation by 6 March.
Activation Partners: Our reflections so far and doubling-down for a second cohort
ARIA’s mission is to unlock scientific and technological breakthroughs that could transform human productivity and wellbeing. But scientific breakthroughs alone are not enough. Our Activation Partner work is designed to create the right conditions for these breakthroughs to be translated into world-changing capabilities.
A year into working with our Activation Partners, ARIA Product Manager Pranay Shah reflects on what we’ve learned from the first cohort — and how the right partnerships can act as force multipliers for long-term impact: read the retrospective.
We’ve also published a concept note for a second call for Activation Partners. Building on the early progress of the first cohort, we want to go further in translating frontier R&D from ARIA’s opportunity spaces into real-world progress through bespoke activities spanning talent, engineering support and organisation building.
We’re also expanding the scope of Activation Partners to apply advanced AI capabilities to R&D, building on insights from our AI Scientists initiative. Share feedback on our concept note.
Benchmarking automated mechanism design: A Q&A with Samuele Marro
To help us shape the development of our Scaling Trust programme, we funded a series of short, exploratory research projects, ranging from exploring aspects of Arena design to diving into topics around physical trust and AI security theory.
One of these projects was led by Samuele Marro and his team, who studied whether agents can automatically design, negotiate and follow game-theoretic interaction protocols. We recently caught up with Samuele to find out more about the team’s project.
Tell us about your ARIA project.
My colleagues (Angelo Huang, Emanuele La Malfa) and I work on negotiation and contract enforcement between LLM agents that don’t trust each other. When humans – or companies – don’t trust each other, they sign contracts. By binding themselves to certain rules, they can then collaborate and build trust without worries about being betrayed. For humans this negotiation process might take months or years, but agents can do it in seconds.
What we showed in our ARIA-funded project is that not only can agents come up with contracts that are useful, but they can also enforce agreements by implementing them as smart contracts. By writing agreements in a programming language and deploying them on blockchains, they can set up enforcement systems that cannot be undermined. This increases trust between agents, makes negotiation much cheaper, and can ultimately lead to more effective agent ecosystems.
What inspired you to work on this specific challenge?
My ‘aha’ moment came when I was testing ways to improve communication efficiency between agents two years ago (back then, it was a way more niche topic). I realised that the simplest solution for increasing efficiency was to let the agents themselves figure out the best approach on their own. From that first result, I started working more on autonomous interactions: if agents could figure out efficient communication, could they figure out precise communication? What about secure communication?
I worked my way up the ladder of questions and realised that all of them were part of a broader goal: enabling coordination between agents. I realised that this research area had a lot of potential, both for improving agent systems and human organisations, and it became one of my main ongoing areas of work.
What do you wish more people knew about your research area?
The number one thing is that agentic negotiation is now surprisingly feasible. In the span of a few years, we went from LLMs barely keeping track of a conversation to agents that can negotiate, implement, and enforce agreements in an end-to-end fashion. The effects of this shift haven’t been fully understood yet, but it will likely unlock use cases that were previously prevented by negotiation costs, such as micro-markets, agent-mediated organisations, and automated supply chains
What do you believe will be the key next steps/applications from your work – how will it help scale trust?
We showed that two agents can negotiate contracts with each other. But why stop there? Why can’t we have 10, 100, 1000 agents negotiate agreements? And why limit ourselves to contracts, when we could design entire collaboration protocols? Our goal is to create an end-to-end system that takes a collection of arbitrary agents and turns it into an effective, trustworthy organisation.
And finally, what book/film/TV show should people check out to understand your project or discipline more?
The Evolution of Cooperation for non-fiction and The Expanse, which is sci-fi, for fiction – the former talks about how cooperation can emerge with a central authority, while the latter is set in a world where three factions have to collaborate without trusting each other (there’s also a TV series!).
Learn more about Scaling Trust.
F-Spec corner: Recommended reads
Our 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-Spec Alice Pettitt has been digging into this month:
Textbooks often display cells as still diagrams, with pathways and structures fixed in place. In reality, cells are constantly in motion, as molecules collide, bind, change shape, and break down at remarkable speed.
This article brings that speed to life through a quantitative metaphor. It takes the opening of a membrane ion channel and imagines slowing it down 10,000 times so that the event lasts as long as a blink of an eye, then uses this comparison to frame other cellular processes. By translating nanosecond molecular events into human-scale experiences, it encourages a shift in focus toward temporal measurement, pushing us to treat time as a central variable for investigation.
A genetically encoded device for transcriptome storage in mammalian cells
How can we move beyond static snapshots to capture the dynamics of gene expression? In this recently published study, the authors introduce TimeVault, an innovative synthetic tool that stores information about gene activity inside living mammalian cells. TimeVault works by capturing messenger RNA (mRNA) molecules inside vault particles, allowing cells to preserve a stable record of which genes were active during a specific window of time.
Remarkably, this record can be retained as cells divide, creating a molecular memory of past cellular states. Unlike other systems, TimeVault does not require scientists to preselect specific genes. Instead, it captures transcriptome-wide information with minimal disruption to the cell. By transforming gene expression analysis from a static snapshot into a form of biological memory, TimeVault opens the door to powerful, time-resolved transcriptomic studies of complex processes such as disease progression.
Adding intrinsically disordered proteins to biological ageing clocks
What if we could detect and prevent age-related disease long before symptoms or irreversible pathology emerge?
While modern medicine can prevent some conditions, it rarely measures the underlying biological decline that precedes age-related disorders. Transforming healthcare will require biological clocks that capture functional deterioration at its source. Beyond epigenetic, telomere, and blood plasma proteome clocks, this perspective proposes a novel intrinsically disordered protein (IDP) clock. Over time, IDPs transition from dynamic conformational ensembles into thermodynamically stable β-sheet-rich amyloid states. Each persistent transition that escapes cellular proteostasis (quality control systems) represents a molecular record of biological ageing.
Supported by findings from model organisms, the hypothesis now requires validation in mammalian systems (e.g., organoids and large longitudinal human cohort studies) and the development of sensitive, non-invasive tools to read the IDP clock in vivo.