Some AI company recently developed some new software so powerful they had to warn and prepare all other major tech companies with special training so their software wouldn't be vulnerable to attacks from the new program. Maybe Anthropic didn't attend this training??