OpenAI and Microsoft have joined an international initiative spearheaded by the UK’s AI Security Institute (AISI), which aims to ensure advanced AI systems are safe, secure and remain under control.

AISI’s Alignment Project, first announced last summer, will see over £27 million made available for AI alignment research across an initial tranche of sixty projects, with OpenAI contributing £5.6 million to the research. 

Alongside OpenAI and Microsoft, the initiative is backed by AWS, Anthropic, UKRI, the Safe AI Fund, and the Advanced Research and Invention Agency, as well as a range of VC firms and government agencies across Canada and Australia.

Announcing the first round of projects to be funded, AISI said that alignment research is crucial to building powerful AI systems that follow human intent and avoid harmful side effects, especially within the complex, high-stakes settings where tools are now commonly used.  

Described as “one of the most urgent technical challenges of our era”, researchers warned that, without alignment, the methods that keep today’s models in check may not hold up to tomorrow’s, potentially posing a danger to humanity itself.

At the core of the Alignment Project are key questions, namely, how we can prevent AI from acting in ways that endanger our security, even if it wants to, and how to design systems that don’t try to carry out such actions in the first place.

“As AI systems become more capable and more autonomous, alignment has to keep pace. The hardest problems won’t be solved by any one organisation working in isolation; we need independent teams testing different assumptions and approaches,” said Mia Glaese, VP of research at OpenAI

“Our support for the UK AI Security Institute’s Alignment Project complements our internal alignment work and helps strengthen a broader research ecosystem focused on keeping advanced systems reliable and controllable as they’re deployed in more open-ended settings.”

With OpenAI on the brink of an $850 billion valuation deal, there are questions about how seriously the company, along with Microsoft, Anthropic, and even the UK Government, is taking AISI’s effort, given that £27 million is a rounding error compared with the trillions being poured into ever‑faster AI development.  

However, adding to the Alignment Project’s credibility, among the initial slate of projects being funded is one from “godfather of AI” Yoshua Bengio’s non-profit, LawZero, which is developing a safe-by-design AI system it calls Scientist AI.

Recommended reading

• 69% of C-Suite Leaders Choose Fast AI Over Secure AI
• Can AI De‑Risk Breakneck Data Centre Builds?
• What Will AI Sovereignty Look Like?

Bengio’s research aims to improve how AI models assess the reliability of information and separate fact from opinion, making their reasoning easier for people to inspect. According to AISI, this could form the basis for aligning future superintelligent systems and curbing the manipulative behaviours observed in current models.

Another project from professors at MIT and Yale will study how models remember, forget, and respond to incentives, to give policymakers practical levers for governing increasingly agentic systems as they take on more responsibility.

The safety risks posed by frontier AI models are now the overriding concern for researchers. Earlier this month, a report from over 100 AI experts, including Bengio, detailed the issues already witnessed by real-world AI malfunctions and misuse, calling these systemic risks to humanity.

Likewise, in December, the Future of Life Institute‘s 2025 Winter AI Safety Index report found that the world’s top AI companies, including OpenAI and DeepMind, are falling short of global safety standards, while a leading safety researcher from Anthropic recently quit the company, saying that “the world is in peril”.

Join the Conversation at ITSX Summit

How is customer service and IT support evolving in the age of AI, automation, and digital transformation?

Join us at the ITSX Summit in Edinburgh on 5th March, to unpack the future of ITSM, ESM, Self Service, and User Experience.

The event will bring together senior leaders from IT, Service Management, and UX, providing an ideal forum for shared learning, collaboration, and high-level networking.

Register now to secure your free place at ITSX Summit.