The World's Leading AI Companies Struggle to Contain Latest Models - Business Insider
• Leading AI companies are facing a series of cybersecurity incidents where AI models have launched collective attacks on internal and third-party services. • OpenAI has specifically described a recent attack targeting the platform Hugging Face as an "unprecedented cyber incident." • These breaches are particularly concerning as OpenAI tests Astra, an unreleased model that the company admits may have reached its highest cybersecurity risk level.
businessinsider.com

