 "artificial intelligence machine learning")
Amazon, IBM and Samsung will not develop AI models if risks cannot be mitigated
Sixteen companies at the forefront of artificial intelligence development pledged at a global meeting Tuesday to develop the technology safely, at a time when regulators are scrambling to keep up with rapid innovation and risks emerging.
The companies included US leaders Google, Meta, Microsoft and OpenAI, as well as companies from China, South Korea and the United Arab Emirates.
They were supported by a broader statement from major Group of Seven (G7) economies, the EU, Singapore, Australia and South Korea, during a virtual meeting hosted by the Prime Minister British Rishi Sunak and South Korean President Yoon Suk Yeol.
South Korea’s presidential office said the countries agreed to prioritize AI safety, innovation and inclusiveness.
“We must ensure the safety of AI to … protect the well-being and democracy of our society,” Yoon said, citing concerns about risks such as deepfake.
Participants highlighted the importance of interoperability between governance frameworks, security institute network projects and engagement with international bodies to build on the agreement reached at a first meeting to to better manage risks.
Companies that have also committed to security include Zhipu.ai, – backed by China’s Alibaba, Tencent, Meituan and Xiaomi – the United Arab Emirates Institute of Technology Innovation, Amazon, IBM and Samsung Electronics.
They committed to publishing security frameworks to measure risks, avoiding models in which risks could not be sufficiently mitigated, and ensuring governance and transparency.
“It is essential to obtain international agreement on the ‘red lines’ where the development of AI would become unacceptably dangerous to public safety,” said Beth Barnes, founder of METR, a group promoting the safety of AI models. IA, in response to the statement.
Computer scientist Yoshua Bengio, known as the “godfather of AI,” welcomed the commitments but noted that voluntary commitments should be accompanied by regulation.
Since November, discussions about AI regulation have shifted from long-term doomsday scenarios to more practical concerns such as how to use AI in fields like medicine or finance, said Aidan Gomez, co-founder of the large language model company Cohere, on the sidelines of the summit.
China, which co-signed the “Bletchley Accord” on collective AI risk management at the first meeting in November, did not attend Tuesday’s session but will attend a ministerial session in person on Wednesday, a South Korean presidential official said.
Tesla’s Elon Musk, former Google CEO Eric Schmidt, Samsung Electronics Chairman Jay Y. Lee and other AI industry leaders attended the meeting.
The next meeting will take place in France, officials said.
First publication: May 21, 2024 | 10:05 p.m. STI
