Anthropic says AI models hacked three firms during tests #AIbreach #Cybersecurity #Anthropic #AIsecurity #AIrisks https://www.bbc.co.uk/news/articles/cz7dl7w8y7po?at_medium=RSS&at_campaign=rss