Newsletter

When Frontier AI Says No to Independent Assessment?

What happens when frontier AI companies can choose whether to submit their most powerful models for independent scrutiny? Responsible Ai UK (RAi UK) latest blog examines Anthropic’s decision not to submit Mythos 5.1 to the UK AI Security Institute, and what it could mean for the future of AI assurance, regulation and public trust.

Read the full blog here.

Is the UK Falling Out of Love with AI?

Is the UK Falling Out of Love with AI?

AI: the growing UK pushback, the latest research briefing from King’s Digital Futures Institute and Responsible Ai UK  (RAi UK) reveals that people are actively pushing back against AI. Surveying over 2,000 UK adults, the findings show usage being deliberately limited, and a generation that can use AI – but is choosing not to. What’s driving the resistance, and what does it mean for responsible AI? Read our full report here.

Responsible Compute Call: AWS cloud compute credits for AI researchers

Responsible Compute Call: AWS cloud compute credits for AI researchers

A new funding call removes a practical barrier that holds back too much responsible AI research: the cost and complexity of accessing high-performance computing at scale.

Responsible Ai UK (RAi UK) has launched the Responsible Compute Call in partnership with Amazon Web Services, funds worth £200,000 in AWS cloud compute credits available to researchers and projects across the programme. Awards of £2,500 or £5,000 in credits give researchers on-demand access to Amazon’s compute instances, machine learning services, scalable storage, and generative AI platforms, without the overhead of managing physical hardware. The call is designed for researchers whose work is currently constrained by access to scalable computing resources. By responsible compute, RAi UK means using these resources efficiently and proportionately, with due regard to the energy and carbon costs of large-scale computation and to the responsible conduct of the research itself.

Round 1 is now open for the researchers who have already received RAi UK funding, or who are currently part of the RAi UK programme. Future rounds will open quarterly and widen eligibility to all postdoctoral researchers and academic staff at UKRI-recognised institutions, until credits are fully allocated. Deadline for Round 1 is 17 July 2026 at 16:00 BST, full details and to apply, visit our funding page.

Six steps to building trust in AI in Healthcare

A new study from Responsible Ai UK Health and Social Care Working Group and the Policy Institute at King’s College London sets out how academic health partnerships and national research communities can support the trusted deployment of AI in the NHS. How much do I need to know about AI in the NHS? Six steps to building trust reports on a policy roundtable that held last month convening fifty representatives from the National Commission into the Regulation of AI in Healthcare, the NHS, academia, industry, policy experts, the voluntary sector, and people with lived experience. Together, they co-produced six trust-building steps:

1. Telling AI’s story from the patient and service-user perspective

2. Building an AI-educated and AI-enabled workforce through peer networks such as the RAi NHS Champions

3. Prioritising AI that tackles health inequalities and treats data quality as foundational

4. Deployment embedding user-centred input across the full AI development pathway

5. Establishing clear guidance on accountability when things go wrong

6. Standardising the assessment and governance of AI tools to reduce variation in adoption decisions.

This is urgent work. The sustainability of health systems globally is severely threatened… supply no longer meets demand and if AI isn’t the solution to the supply side of the equation, then what is?

Another study from RAi UK , King’s Health Partners and the Policy Institute at KCL ‘The use of AI in UK Healthcare’ finds one in seven people have used AI instead of seeing a GP. The public are turning to AI chatbots for health advice, and many say it has shaped their decisions about whether to seek professional care. Read the full study and our recommendations.

RAi UK International Sandpit at Singapore Advances Global AI Assurance

April 25-28, 2026, Singapore: Responsible AI UK co-hosted its second international Research Sandpit with the NExT++ Research Center at the National University of Singapore, bringing together 18 researchers from the UK and Singapore to address a shared challenge: advancing responsible AI assurance across diverse cultural and regulatory contexts. Led by Sandpit Directors Prof Derek McAuley (University of Nottingham) and Prof Kate Devlin (King’s College London), and guided by mentors Prof Jack Stilgoe (UCL), Prof G. Owen Schaefer and Prof Jungpil Hahn (NUS), the week was rich with challenge, rigour, and possibility. Compelling provocations came from Sebastian Porsdam Mann (NUS), Sang Hao Chung (IMDA) and Chris Barnes (NPL) and expert facilitators Toby Scott and David Lomas shaped a programme that moved fluidly from community-building to deep intellectual exploration. The week closed with team presentations, peer feedback, and a funding clinic. With contributions from Prof Joel Fisher and anchored by Prof Dame Wendy Hall’s powerful call for long-term research, shared datasets, and a common language to support global AI governance.

RAi UK is  deeply grateful to Deputy High Commissioner Lucy Hughes and the British High Commission Singapore for so warmly hosting our delegation at their official residence, a great opportunity for conversations to strengthen UK–Singapore collaboration. The evening underscored the importance of building trusted research ecosystems through sustained international partnership. This is another landmark step in our mission to advance AI assurance across cultures and borders, to know more about our research programme, visit our website.