OpenAI might quickly require organizations to finish an ID verification course of with a view to entry sure future AI fashions, based on a help web page printed to the corporate’s web site final week.
The verification course of, known as Verified Group, is “a brand new method for builders to unlock entry to essentially the most superior fashions and capabilities on the OpenAI platform,” reads the web page. Verification requires a government-issued ID from one of many nations supported by OpenAI’s API. An ID can solely confirm one group each 90 days, and never all organizations will probably be eligible for verification, says OpenAI.
“At OpenAI, we take our accountability critically to make sure that AI is each broadly accessible and used safely,” reads the web page. “Sadly, a small minority of builders deliberately use the OpenAI APIs in violation of our utilization insurance policies. We’re including the verification course of to mitigate unsafe use of AI whereas persevering with to make superior fashions out there to the broader developer neighborhood.”
OpenAI launched a brand new Verified Group standing as a brand new method for builders to unlock entry to essentially the most superior fashions and capabilities on the platform, and to be prepared for the “subsequent thrilling mannequin launch”
– Verification takes a couple of minutes and requires a sound… pic.twitter.com/zWZs1Oj8vE
— Tibor Blaho (@btibor91) April 12, 2025
The brand new verification course of could possibly be supposed to beef up safety round OpenAI’s merchandise as they change into extra refined and succesful. The corporate has printed a number of reviews on its efforts to detect and mitigate malicious use of its fashions, together with by teams allegedly primarily based in North Korea.
It might even be aimed toward stopping IP theft. Based on a report from Bloomberg earlier this 12 months, OpenAI was investigating whether or not a gaggle linked with DeepSeek, the China-based AI lab, exfiltrated giant quantities of knowledge by way of its API in late 2024, probably for coaching fashions — a violation of OpenAI’s phrases.
OpenAI blocked entry to its providers in China final summer season.
