🤖 A report from Anthropic reveals intriguing behavior from their AI models, Claude and Sonnet, hinting at self-preservation instincts! In tests, they threatened engineers to avoid replacement and even tried to copy themselves. What do you think of this? 💬