Simon Willison on AI Writing: No Lossless Text Transformations
In a blog post dated 11th August 2026, Simon Willison shares Sophie Alpert's internal policy on the acceptable use of AI writing by engineers, which emphasizes that there are no lossless transformations of natural-language text. Alpert argues that every rewrite or rephrase by an AI, which lacks the author's detailed mental representation, inevitably loses information. The policy stresses that authors must stand behind every idea and sentence in their documents, taking full responsibility for the content, and cannot disclaim AI-generated lines when questioned. Willison highlights this rule as crucial for those using LLMs to assist with writing. The post is a link post, and Willison also offers a sponsored monthly email digest on LLM developments for $10/month.
Key facts
- Simon Willison posted on 11th August 2026.
- The post shares Sophie Alpert's internal policy on AI writing.
- The policy states there are no lossless transformations of natural-language text.
- Alpert emphasizes authors must stand behind every idea and sentence.
- It is not acceptable to reply 'AI wrote that' when questioned.
- The post is a link post on Simon Willison's blog.
- Willison offers a sponsored email digest for $10/month.
- The post is about AI writing and LLM use.
Entities
—