OpenAI, Anthropic and GoogleDeepMind are working on creating an oversight body for the risks associated with artificial intelligence (AI), OpenAI said.
The idea of a supervisory entity arose from a proposal by Demis Hassabis, of Google DeepMind, to create a body inspired by the Financial Industry Regulatory Authority (Finra), which supervises brokers and investment firms in the United States.
“Funding would have to be substantial and would likely come mostly from industry,” Hassabis wrote.
Likewise, Sam Altman, CEO of OpenAI, creator of ChatGPT, expressed his support for collaboration across the sector and acknowledged public fear that competition between companies and countries could put security on the back burner.
Progress slowdown
“I think it’s great that our industry is saying we want to come together, we want to be able to coordinate and make sure we have enough time to do this safely,” Altman said at a conference.
Once the oversight structure is in place, the most advanced companies in AI should subject their new models to a review, focusing in particular on security, before they can be marketed.
This new body could also push for a coordinated slowdown in AI development, should risk understanding not keep pace with technology progress.
Refusal to suspend
For his part, Meta’s Mark Zuckerberg rejected growing calls to suspend or slow down the development of AI, considering that market mechanisms and the risk of lawsuits constitute the best counterweights to this technology.
“People will not want to use agents that are not aligned with their interests and do not do what they are asked. Labs therefore have a strong natural incentive to make their models more aligned,” Zuckerberg wrote in X.
“Alignment” refers to efforts to get models to behave the way humans want them to. “Any laboratory that does not focus on alignment will be left behind,” he added.
Meanwhile, Canada and Germany intend to “transform research into reliable AI products” in their “commitment to democracy and values,” said German Minister of Digitalization, Karsten Wildberger.
We are at a “crucial moment” in the development of AI, and Berlin and Ottawa want to support technology “at the service of the population,” said Evan Solomon, Canada’s digital officer.