AI Disclosure

How Alto Uses AI

Plenty of sites now quietly generate their content. Alto does not, and this page exists so you can hold us to that — here is every place AI touches the site, and every place it does not.

Where AI appears

There are three places, and only three.

  • Story analysis. Many story pages carry a Gab AI Analysis block beneath the article: a short read on what the story means, what context is missing, and where it fits in a larger pattern. It is generated on request or from a cached earlier run, and it is always labeled.
  • Related-source lookup. To assemble the “other coverage” list next to an analysis, Alto uses a search-grounded model that returns real URLs. Those links point at the publishers themselves, so you can check the claim rather than take the summary's word for it.
  • Editorial copy on hub pages. Some programmatic hub pages — a market ticker, a state weather page — have introductory copy that was drafted with model assistance and then reviewed before publication. This copy is descriptive and evergreen; it never asserts a live figure. Live numbers on those pages always come from the data source named on the page.

Where AI does not appear

  • Headlines and articles. Alto does not write, rewrite, or paraphrase-and-republish anyone's reporting. What you read as the story is the publisher's own text, and the link goes to their site.
  • Rankings. Trending and Most Read are arithmetic on read counts. No model decides what you see first.
  • Tool results. Every calculator and lookup on Alto is deterministic code over official data. A tax bracket, a mortgage payment, a lottery match, a bill status — none of it is generated. Run the same inputs twice and you get the same answer.
  • Source selection. Which publishers Alto monitors is a human decision, documented in the source directory.

Which models, and who runs them

Analysis runs through Gab AI on top of open-weight and commercial models accessed via a model gateway. Alto uses a primary model with an automatic fallback, so a provider outage degrades to a second model rather than to a blank page. The specific model is a configuration value and changes as better ones ship; the disclosure, the labeling, and the limits described here do not change with it.

What is sent to the model is the article's title, text, and publisher. No reader identity is attached, because Alto does not have one to attach.

Limits you should assume

Treat AI analysis the way you would treat a well-read acquaintance summarizing an article they skimmed. It is often useful and sometimes wrong. Specifically, it can misattribute a quote, merge two similar events, get a date or a number wrong, or state a contested claim as settled. It has no ability to verify anything, and a confident tone is not evidence.

The original article is always the authority. That is why every story page links to it prominently, above the analysis rather than below it.

Perspective, stated up front

Alto's analysis is written from a right-leaning, Christian perspective. We think disclosing a point of view is more honest than performing neutrality, and it is why the analysis sits in a labeled block instead of being woven into the summary. If you want the story without the commentary, read the publisher's version — the link is right there.

Reporting an error

If an AI analysis on Alto is wrong, tell us and we will correct or remove it. See Corrections. For how AI fits into Alto's broader policies, see Editorial Standards, and for how the rest of the site is built, see Methodology.

Frequently Asked Questions

Is any of the news on Alto written by AI?
No. Every article on Alto was written and published by a third-party newsroom, and each one credits that publisher and links to the original. AI on Alto only adds an optional analysis layer beneath the story; it never generates a headline or a report.
How do I tell AI-generated text from the publisher's reporting?
AI output is confined to a labeled block titled Gab AI Analysis, visually separated from the article. If text is not inside that block, it came from the publisher, from an official data source, or from Alto's own editorial pages.
Can Alto's AI analysis be wrong?
Yes. Language models can misread a source, conflate similar events, or state something confidently that is not true. The analysis is context, not verification. Read the original article before relying on any claim, and report errors through the corrections page.
Does the AI analysis have a political point of view?
Yes, and Alto says so plainly. The analysis is written from a right-leaning, Christian perspective. It is commentary offered as commentary, not a claim of neutrality.
Is my reading used to train an AI model?
No. Alto requires no account and builds no reader profile. What is sent to the model is the article text and its metadata — not who is reading it.