Updated April 2nd, 2024 at 17:58 IST

US, Britain to collaborate on improving AI safety mechanisms

Britain and the United States intend to conduct at least one collaborative testing exercise involving a publicly accessible AI model.

Reported by: Business Desk
Artificial intelligence | Image:Pexels

US-UK partnership: The United States and Britain have announced a new collaborative partnership focused on improving the safety of artificial intelligence (AI), addressing mounting concerns surrounding forthcoming iterations of the technology.

Commerce Secretary Gina Raimondo and British Technology Secretary Michelle Donelan formalised this agreement by signing a memorandum of understanding in Washington. The pact aims to jointly advance the testing of advanced AI models, building upon prior commitments made at an AI Safety Summit held at Bletchley Park in November.


Raimondo highlighted the role of AI in shaping contemporary technological landscapes, stating, "We all recognise AI as the defining technology of our era." She underscored that this partnership will expedite the efforts of both nations in mitigating the risks associated with AI, encompassing national security concerns as well as broader societal impacts.

Both Britain and the United States have established government-led AI safety institutes. Britain disclosed plans in October for its institute to scrutinise and evaluate novel AI advancements, while the United States announced in November the establishment of its own safety institute dedicated to assessing risks posed by cutting-edge AI models. Currently, the US initiative involves collaboration with 200 companies and entities.


Through this formalised partnership, Britain and the United States intend to conduct at least one collaborative testing exercise involving a publicly accessible AI model. They are also exploring the possibility of personnel exchanges between their respective institutes. Moreover, both nations are actively seeking to forge similar partnerships with other countries to foster global collaboration on AI safety.

Donelan highlighted the role of this agreement, stating, "This marks the first of its kind anywhere in the world." She acknowledged the transformative potential of AI in addressing global challenges but stressed the imperative of effectively managing associated risks.


Generative AI, which possesses the capability to generate text, images, and videos in response to open-ended prompts, has garnered both excitement and apprehension. Concerns have been raised regarding its potential to render certain jobs obsolete, influence electoral processes, and potentially surpass human capabilities with potentially catastrophic consequences.

In a joint interview with Reuters, Raimondo and Donelan underscored the urgency of concerted action to address AI risks. Donelan highlighted the imminent release of significantly more capable AI models, necessitating swift and targeted interventions.


Raimondo announced plans to address AI-related concerns at an upcoming meeting of the US-EU Trade and Technology Council in Belgium. She also revealed the Biden administration's intention to bolster its AI team.

Both countries have committed to sharing crucial information regarding the capabilities and risks associated with AI models and systems, as well as collaborating on technical research pertaining to AI safety and security.


In October, President Biden signed an executive order aimed at mitigating AI risks, while in January, the Commerce Department proposed regulations requiring US cloud companies to ascertain whether foreign entities are accessing US data centres to train AI models. Meanwhile, Britain announced plans in February to allocate over 100 million pounds ($125.5 million) towards the launch of nine new research hubs and initiatives aimed at educating regulators about AI technology.

Raimondo expressed particular concern regarding the potential misuse of AI for bioterrorism or nuclear war simulations, stressing the necessity of a zero-tolerance approach towards such capabilities.




Published April 2nd, 2024 at 17:58 IST