It’s a break with the past.
In May 2023, meeting in Hiroshima, Japan, G7 leaders launched the so-called Hiroshima AI process to promote safe and trustworthy AI and signed a statement that explicitly recognized “the need to manage risks” and pushed for “keeping humankind at the center.”
It came with a voluntary code of conduct for companies developing the most advanced AI models. That code of conduct was namechecked only once in the current G7 draft.
The safety focus back then followed the overnight popularity of generative AI thanks to OpenAI’s ChatGPT.
In the meantime, however, attention has shifted from safety to winning the AI race.
In January, U.S. President Donald Trump revoked an executive order from the Biden era that had aimed to develop “safe, secure and trustworthy” AI, in a rare win for the EU.
