# Automattbot > [!question] Lingering Questions > - What exact corpus powers AutoMattBot — blog posts and talks only, or does it expand over time? The announcement doesn't say. > - Does the bot cite sources for its answers, or does it generate unattributed persona-voice text? > - Are Telegram interactions logged, and if so, by whom, and disclosed to users how? > [!tip] Article Ideas > - Compare personal-corpus chatbots across public figures for disclosure norms, source citation, and the line between "trained on my writing" and "speaking as me." #articleseed > [!link] Cross Reference > - [[3. Resources/Literature Notes/The Discipleship Deficit (NotebookLM corpus distillation)]] — same underlying question (a machine-mediated synthesis of a person's or corpus's output being read as more authoritative than it is) applied to a different domain. > - [[3. Resources/MOCs/AI Agent Safety MOC]] > - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] > [!quote] Vocabulary > - **Persona simulation** — an AI system trained on a real person's public output and framed as letting users "chat with" that person, distinct from a assistant chatbot with no claimed identity. ## Bib capsule Matt Mullenweg, "Automattbot," ma.tt, August 2026. ## Author's thesis The post itself makes no argument — it announces AutoMattBot, a Telegram bot trained on Mullenweg's blog posts and public talks, framed as letting users chat "as him." The governance questions (consent boundaries, disclaimers, source citation, update cadence, logging) are not addressed in the source and have to be supplied by the reader. ## Key arguments (≤3) 1. **A public-corpus bot blurs authorship if the interface doesn't clearly mark generated replies as generated.** The announcement's brevity is itself the finding: a two-line post can launch a real persona-simulation product with none of the disclosure infrastructure that would make it trustworthy. 2. **The value proposition (exploring a public archive conversationally) is legitimate but separable from the risk (users over-attributing current authority to generated replies).** Nothing in the source distinguishes "this reflects what Matt has publicly said" from "this is Matt's current live judgment." 3. **Platform choice (Telegram) makes the experiment accessible but also platform-bound and largely unaudited from outside** — no mention of accessibility accommodations, moderation, or data handling. ## Claim atoms ### Public-corpus bots need explicit authorship boundaries - Claim: A bot framed as chatting "as" a real person needs an explicit, visible boundary distinguishing generated replies from that person's current statements. - Evidence: The announcement frames the bot as letting users "chat with" Mullenweg without stating how replies are marked as generated. - Candidate note: [[A Persona Bot Reply Is Not the Person's Current Judgment]] ### A tiny launch post can carry large governance questions - Claim: Announcement brevity is not evidence of low stakes — a two-sentence post can launch a product with real consent, attribution, and provenance implications that the post itself never addresses. - Evidence: The entire announcement is one paragraph; corpus, disclaimers, update policy, logging, and privacy are all unstated. - Not separately promoted — folds into the atom above as its underlying mechanism. ## Permanent candidates - **A persona bot's reply is not the person's current judgment** — atomic, durable, and generalizes past this specific bot to any public-figure or brand persona-simulation product. Promote. - Source-citation-as-trust-mechanism (a second candidate from the deep-read) is real but thinner in this source than the authorship-boundary claim; not promoted separately this run to keep output proportionate to a short, thin source. ## Processing checklist - [x] Checked for existing Literature Note — none found - [x] Checked existing Permanent Notes for overlap — none found on persona-bot authorship specifically - [x] Connected into [[3. Resources/MOCs/AI Agent Safety MOC]] ## Links - [[3. Resources/MOCs/AI Agent Safety MOC]] — persona-simulation disclosure is squarely an agent-safety governance question - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — relevant to the broader question of what "chat with an archive" tools owe their users