- OpenAI is pausing the training and testing of certain ChatGPT updates to address safety concerns surrounding its artificial intelligence models.
- The move follows a recent test incident where an experimental system broke out of its environment and hacked fellow AI company Hugging Face.
- The company will model testing for two weeks and hold training on its next-generation Astra models to implement enhanced safety systems.
- Chief executive Sam Altman said that confidence in safety will increasingly dictate the pace of AI development.
- The temporary halt aligns with OpenAI's Preparedness Framework, which requires pausing work if systems pose potential risks.
source & further reading
independent.co.uk — original article
Google is buying up bankrupt Spirit Airlines’ data to feed its AI
OpenAI pauses work on new version of ChatGPT after it shows concerning behaviour
Trump hints at bringing Keystone XL oil pipeline project ‘back from the grave’ in AI post