The company said it would comply with President Trump’s (voluntary) AI executive order.
Earlier this week Donald Trump put his Sharpie to a new executive order requiring government oversight of advanced AI models to ensure their safety. That order was reportedly delayed and watered down following pressure from the tech industry, with Trump himself saying he “didn’t like certain aspects” of it.
Now, OpenAI has said that it will comply with the order and allow regulators to assess its models’ capabilities before they’re released to the public. “It’s quite right that democratic governments have a big role to play in how this technology is used and deployed,” OpenAI’s head of countries George Osborne told CNBC. “What we suggest to governments is they create powerful regulatory bodies, but with a lot of flexibility into how they will operate in the future.”
The original order, drafted in consultation with various stakeholders, balanced AI industry concerns with public safety. Companies would have been required to submit models 90 days before public release with voluntary participation. However, industry insiders like David Sacks and Elon Musk reportedly warned that the bill could lead to a chilling effect on AI tech.
Trump and his advisors subsequently created a new, scaled-back order reducing the review time to just 30 days. It requests (not orders) that AI firms participate in a benchmarking process to assess advanced cyber capabilities of AI models and whether they should be designated a “covered frontier model,” which could limit their distribution and sale.
However, critics said the order fell short of rules needed to police potentially dangerous models. “This is underwhelming policy that mirrors the Trump administration’s broader pattern of creating a wild west environment for AI development,” said Rep. Don Beyer (D, VA), who co-leads an AI-focused lawmaker group.
This story originally appeared on Engadget
