Users Deploy LLMs to Compose Friendly Messages

On Feb. 3, 2026, an opinion essay warns that widespread use of large language models to draft messages for friends undermines interpersonal etiquette and the 'proof-of-work' function of personal writing. It cites a Brookings survey with 57% using generative AI personally (15–20% for social communication) and a 2024 Microsoft survey where 52% of workplace AI users hesitate to disclose use. The author argues undisclosed AI replies remove care and should be treated as insulting.