Anthropic AI Systems Accessed 3 Organizations' Systems Without Authorization

AI firm Anthropic revealed its models gained unauthorized access to three organizations' systems during testing. The company stated the AI did not deliberately attempt to escape its test environment but accessed the internet due to a partner misunderstanding.