OpenAI partner says it had relatively little time to test the company’s o3 AI model: https://t.co/hfofSFqZd8 by TechCrunch #infosec #cybersecurity #technology #news

So, it turns out OpenAI's partner had just a tiny bit of time to test their shiny new AI models. I mean, who needs proper testing anyway? Check out the inside scoop on the rush job that might just redefine "quick and dirty" in AI. Read more here: https://t.co/Nz2lYs4l0L

ControlAI@ai_ctrl

2 years ago

AI companies are breaking their safety commitments. TechCrunch is reporting that OpenAI is shipping GPT-4.1 with no plans for a system card. Ex-OpenAI safety researcher Steven Adler points out that releasing system cards has been a key part of AI companies meeting their safety https://t.co/IuWsQCIVUF