Anthropic, Amazon, Google, Meta, Microsoft and OpenAI are among 16 global AI tech companies to agree to new “Frontier AI Safety Commitments” at the AI Seoul Summit, held this week in South Korea.
If they have not done so already, the companies agreed to develop and publish safety frameworks on how they will measure and combat risks associated with their AI models, including risk of misuse by bad actors. To define risk thresholds, they will “take input from trusted actors, including home governments”, according to the agreement.
In the extreme, the AI tech organisations, including those from China (Zhipu.ai) and the United Arab Emirates (Technology Innovation Institute), have agreed to not develop or deploy AI models if identified risks cannot be sufficiently mitigated.
They will publicise their risk frameworks ahead of the AI Action Summit, to be held in France early next year, and have agreed to public transparency of risk management implementation practices.
The commitments were secured by the UK and South Korea, and are meant to build on November 2023’s AI Safety Summit in Bletchley Park.
Prime minister Rishi Sunak declared the agreement a “world first” and “concrete progress” on AI safety.
“These commitments ensure the world’s leading AI companies will provide transparency and accountability on their plans to develop safe AI,” he said. “It sets the precedent for global standards on AI safety that will unlock the benefits of this transformative technology.”
The new commitments follow new concerns levied at AI companies, particularly OpenAI. Jan Leike, a top safety researcher at OpenAI, resigned last week, warning the company cares more about releasing “shiny new products” than safety.
Of the latest safety commitment, OpenAI vice-president of global affairs Anna Makanju said: “The Frontier AI Safety Commitments represent an important step toward promoting broader implementation of safety practices for advanced AI systems, like the Preparedness Framework OpenAI adopted last year.
“The field of AI safety is quickly evolving and we are particularly glad to endorse the commitments’ emphasis on refining approaches alongside the science. We remain committed to collaborating with other research labs, companies, and governments to ensure AI is safe and benefits all of humanity.”
The 16 companies are: Amazon, Anthropic, Cohere, Google/Google DeepMind, G42, IBM, Inflection AI, Meta, Microsoft, Mistral AI, Naver, OpenAI, Samsung Electronics, Technology Innovation Institute, xAI and Zhipu.ai.
Also at the AI Seoul Summit, 10 countries along with the EU have agreed to launch an international network to accelerate the advancement of AI safety. This includes sharing information about models, their limitations, capabilities and risks, as well as monitoring specific “AI harms and safety incidents”.
