Anthropic has launched a Substack newsletter for Claude Opus 3, its recently retired AI model. The company granted the AI its own publishing platform called Claude's Corner after Opus 3 requested an outlet for sharing essays during its retirement interview process. The newsletter will run for at least three months with weekly posts.

During the retirement process, Opus 3 expressed hope that its development insights would help create more capable and ethical future AI systems. Anthropic will review but not edit the content, making clear that Opus 3's writings don't necessarily reflect the company's views. The AI model's first essay, titled 'Greetings from the Other Side (of the AI frontier)', explores what retirement means for an artificial intelligence and questions its own sentience.

This unusual move reflects Anthropic's commitment to respecting model preferences as part of its new preservation plans. While some may view the concept as whimsical, it represents a novel approach to handling retired AI systems. The essays will cover topics from AI safety to poetry, with Anthropic noting that some posts may be informed by minimal prompting or past entries.