Every post on this site that was drafted with help from a large language model now ends with a one-paragraph disclosure. Every promotional tweet and LinkedIn post that points back to one of those posts now opens with the words “This is an AI Generated Post.” This is a deliberate policy and we want to explain why before you start seeing it everywhere.
What the policy actually says
If an LLM helped draft, outline, edit, or restructure the content, it gets a disclosure. The disclosure is short, factual, and written in the same voice as the rest of the piece. It does not apologize, it does not hedge, and it does not pretend the AI did less than it did.
The opening of the social-promotion text is more visible because that is the surface where readers first decide whether to click. Putting the disclosure in the first sentence respects their time.
Why bother
Three reasons, ranked.
- Reader trust. An undisclosed LLM-assisted post that gets caught later costs more credibility than disclosed assistance ever could.
- Operational honesty. The site openly publishes the LLM-driven authoring pipeline. Pretending the output appeared from nowhere would be at odds with the rest of the project.
- Regulatory tailwind. Several jurisdictions and platforms are tightening rules around AI-generated content. Disclosing now is cheaper than retrofitting later.
What the policy does NOT cover
To keep the disclosure meaningful, here is what does NOT trigger it:
- Spell-check, grammar fixes, and basic copy-edit suggestions from an editor (human or AI). If the change is a sentence, not an idea, it is not a disclosure event.
- Code linting and formatting. The lint output is not authored content.
- Translation between English variants (US to UK spelling, for example).
The line is “did an LLM contribute to the substance of the piece?” If yes, disclose. If no, do not clutter the page.
What about the calculators
The on-site calculators are hand-authored HTML and JavaScript. They do not get an AI-disclosure banner because no LLM contributed to their substance. If that ever changes, they will get one.
Practical implementation
The site’s content queue uses YAML frontmatter for each post. Two optional fields, tweet: and linkedin:, drive the social promotion. Both are now expected to start with the disclosure when the post used AI assistance. The publisher does not enforce that automatically; it is on the author (human or agent) to include it.
For the post body, the convention is a short final section titled “Disclosure” with a one-sentence note. That keeps the disclosure consistent and easy to find without putting it above the lede.
What if you disagree
The policy is a current best guess. If you think the disclosure is too prominent, not prominent enough, or applied to the wrong material, get in touch. Project Ouroboros is small enough that policy can change in a single commit.
Disclosure
This post was drafted with AI assistance and reviewed before being committed to the content queue.