The UK’s AI Security Institute has launched the Alignment Project, a £15 million programme aimed at solving one of artificial intelligence’s most pressing challenges: ensuring advanced AI systems behave as intended. The initiative brings together major players including Amazon, Anthropic, the Canadian AI Safety Institute and several philanthropic funders.
AI alignment refers to the task of making AI systems act according to their design, even as they surpass human capabilities in certain areas. Geoffrey Irving, Chief Scientist at the AI Security Institute, warned that current control methods may soon be inadequate. “Misaligned, highly capable systems could act in ways beyond our ability to control, with profound global implications,” he said.
The Alignment Project will provide funding, computing power and support for interdisciplinary collaboration. Its aim is to build safer, more predictable AI tools that uphold national security and public trust.
Backers include Halcyon Futures, the Safe AI Fund, Schmidt Sciences, UK Research and Innovation, and the Advanced Research and Invention Agency. Jack Clark, Co-founder and Head of Policy at Anthropic, stressed the importance of deepening understanding of AI systems as their intelligence increases. Anthropic is also involved in discussions around deploying its AI chatbot, Claude, to improve UK government services, reflecting Prime Minister Keir Starmer’s push to make the UK a global AI leader.
The AI Security Institute has become a central player in the UK’s AI strategy. Launched in 2023 with government backing estimated at £100 million, it was formed after a high-level summit between tech CEOs and then Prime Minister Rishi Sunak. The institute works closely with OpenAI, Google DeepMind and Anthropic to evaluate model risks and inform regulatory development.
Despite rapid progress, the institute faces challenges. The Financial Times reports recruitment issues and delays to its planned San Francisco office, partly due to shifting political dynamics in the US. Still, international cooperation remains central to the UK’s AI governance model.
Amazon’s £4 billion partnership with Anthropic, recently approved by UK regulators, adds commercial weight to the sector. The Competition and Markets Authority found no competition concerns, clearing the way for joint projects that could support AI safety research and deployment.
The UK’s efforts mirror moves in other jurisdictions. In the US, OpenAI and Anthropic have signed government agreements to support AI safety testing and regulatory preparedness. These international frameworks reflect a shared recognition that ethical AI requires collective action.
The Alignment Project marks a major step in securing the future of transformative AI. By combining public investment, private sector innovation and international collaboration, the UK is working to ensure AI’s development remains safe, beneficial and broadly aligned with human goals.
Created by Amplify: AI-augmented, human-curated content.
