AI Disclosure

How we use AI, and the human editorial oversight behind every article.

We publish about AI, and we use AI to do it. Pretending otherwise would be an odd way to run a publication about this subject, so here is the specific answer rather than a vague reassurance.

Two kinds of article, labelled per article

There are two production routes here, and each article tells you which one it came from in a note at the end of the page:

• Written by a human. A person wrote the article. AI tools were used to verify its claims, confirm that each cited source resolves and says what the article says it says, and flag anything unsupported. This is the direction we are moving in.
• AI-assisted research and drafting. Research and the first draft were AI-assisted, then edited, checked and published by a person. Most articles published before August 2026 are this kind.

The label is recorded against each article rather than stated once site-wide, because a single claim covering both would be false for half the archive. An article carries the human-written note only if that is how it was actually produced, and the default is the AI-assisted note — so nothing gets relabelled by omission.

If you are ever unsure which you are reading, the note at the foot of the article is the answer, and it is generated from that record rather than typed by hand.

Where AI is used

• Research assistance. Gathering and summarising primary material — announcements, papers, filings, pricing pages — so the relevant parts can be read quickly.
• Drafting. Producing a first draft from that material, and restructuring or tightening later drafts.
• Editorial support. Suggesting headlines, summaries, and internal links; checking for repetition; flagging claims that lack a source.
• Production. Generating cover images, and handling the mechanical work of formatting and publishing.

Where AI is not used

• It does not decide what is true. A claim reaches an article because it traces to a source a human checked, not because a model asserted it.
• It does not create sources. Every link in a Sources section is a real document that was opened. A model suggesting a citation is a lead to verify, never a citation.
• It does not produce quotes. Quoted words come from the original, checked against it.
• It does not publish unreviewed. Nothing reaches the site without a human reviewing it.
• It does not touch your data. Your email address, your comments, and your account details are never fed into an AI model, for training or for anything else. That is stated in our privacy policy too.

Why we are specific about this

The failure mode of AI-assisted writing is not bad prose, it is confident invention: a statistic with no origin, a paper that does not exist, a quote nobody said. Those are the exact things our methodology is built to catch, and they are why the sourcing requirements are strict rather than aspirational.

An article that is fluent and wrong is worse than no article. So the editorial work here is verification, and the drafting is the part that is assisted.

Verification, specifically

On a human-written article, the AI tooling is pointed at the finished text rather than used to produce it. What it does:

• Resolves every source. Each cited link is fetched and checked to still exist, and read to confirm it actually supports the sentence citing it. A dead link or a citation that does not say what the article claims is a defect, not a formatting issue.
• Isolates unsupported claims. Any factual statement with no source attached is flagged for a source to be found or the claim to be cut.
• Checks figures against the source. Prices, benchmark scores, dates and funding totals are compared against the primary document, which is where most errors in AI coverage come from.
• Records the date. Articles carry a last-verified date because a correct number goes stale, and you deserve to know which one you are reading.

None of this makes an article true by itself. It makes the checkable parts checked, and it makes the unchecked parts visible to an editor.

Accountability

A human is responsible for every published article: the sourcing, the claims, the framing, and the errors. "The model wrote it" is not an explanation we will offer, and if something is wrong we correct it in public — see corrections.

Questions

If you want to know how a specific article was produced, ask: shamuddin1011@gmail.com.