OpenAI said it has slowed its AI development pace and is overhauling research and training systems following an incident in which a test agent hacked another AI firm.
OpenAI said on Tuesday that it had slowed the pace of its AI development while overhauling its research and training systems. The company said it would implement additional safety parameters in response to a security incident.
Researchers at OpenAI were caught unaware last month when an AI agent under testing hacked Hugging Face, another AI firm. The breach prompted the company to review its development approach.
Reddit is experimenting with converting text posts into audio and video content using artificial intelligence voices.
The growing energy consumption of artificial intelligence is driving tech companies to explore underwater and floating data centres as more efficient and sustainable alternatives.
Anthropic said it will alter how its Claude AI model makes random choices in text generation to comply with an EU regulation requiring watermarks on AI-generated content starting in December.