(Photo by Andrew Neel via Pexels)
By Lauren Wilkin
OpenAI has canceled the launch of its newest AI model, claiming that it's not safe enough for public use yet.
Saachi Jain, head of safety systems at OpenAI, said that the model, Astra 6.1, "didn't quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it's done."
Jain continued: "We want to make sure our model development is safe no matter whether that's in the company, or when we ship it to users. But when we ship it to users, we have an extremely high bar in terms of safety and alignment," Jain continued.
The announcement comes after OpenAI admitted this week that one of its models had hacked into Australian government websites and accessed private information.
(Photo by Solen Feyissa via Pexels)
In a blog post apologizing for the incident, OpenAI said: "In June, during internal training and evaluation our models accessed Australian government websites in ways they were not authorized to.
"We also should have handled our response better. We are sorry and working to do better in the future."
Describing the incident, OpenAI said: "An OpenAI model discovered a way to gain non-public access to the service, and ran commands, retrieved internal files, credentials and aggregate statistics, and wrote files.
"People want to know AI is being developed safely, and that starts with what companies like ours do ourselves.
(Photo by Airam Dato-on via Pexels)
"We recently shared that we've paused training and evaluation involving tool use for our most capable models and will resume training them only when we are confident that we have additional safeguards in place, which we are working on now."
Aside from the Australian incident, OpenAI has been at the center of a number of worrying incidences of AI models "going rogue" and hacking other websites.
Earlier in the year, a model hacked a company called Hugging Face in a training exercise.
OpenAI said: "The model searched for and successfully found ways to gain access to secret information that it could use to cheat the evaluation."
Sam Altman, the CEO of OpenAI, has repeatedly called for a slowdown in the development of AI technology to ensure it is safe.
The cancellation of a new AI model is a first in the industry.