How we work — editorial standards, sourcing, and corrections.

1. Editorial principles

Four principles guide every piece we publish:

1. Accuracy over speed. If a claim isn't verifiable, it doesn't run. If two sources disagree, we name both and explain the disagreement.
2. Independence from vendors. No paid placements. No advertiser preview of editorial. No "embargo in exchange for favorable coverage."
3. Transparency. We disclose affiliate relationships, show our testing methodology, and cite original sources.
4. Usefulness. We write for operators — builders, indie devs, marketing and ops teams using AI to get work done. "Interesting" is not enough; a reader should finish a piece knowing what to do differently.

2. Sourcing & citation

Every factual claim in a news piece or review links to at least one primary source. Primary sources include:

• Official AI lab blogs and release notes (e.g. Anthropic, OpenAI, Google DeepMind, Meta AI Research, Microsoft Research, NVIDIA Research).
• Peer-reviewed research and pre-prints on arXiv, with author names and institutions.
• Official government or regulatory announcements (e.g. FTC, EU AI Office).
• Company SEC filings for public-company earnings coverage.
• First-hand testing we performed ourselves, with screenshots or reproducible prompts where possible.

Secondary sources (other news outlets) are used only to corroborate primary sources, never as the sole source of a factual claim. When we reference another outlet's original reporting, we credit them by name and link to the piece.

3. Review methodology

For product reviews and comparison articles, we follow a consistent process:

1. Define the category. We publish our evaluation criteria before testing (speed, accuracy, price, ecosystem, etc.) so scoring is transparent.
2. Hands-on testing. We use each product for real tasks. For coding tools, that means shipping actual code. For image generators, running the same prompt battery across all contenders. For chatbots, testing the same question set.
3. Scoring. We score on a 5-point scale per criterion. Weighted totals drive the ranking. Scoring sheets are available on request.
4. Verdict. We pick a category winner, a "best value" pick, and — where warranted — a "runner-up." Every pick includes our specific reasoning.
5. Disclosure. We state upfront which products provided a free trial or review credits, and whether we have an affiliate relationship.
6. Revisit cadence. Reviews are dated. We revisit category winners quarterly or whenever a major version ships.

We do not accept payment for reviews, for higher rankings, or for inclusion in comparison tables. Products we review may, separately, run ads on our site; those placements are labeled as sponsored and have no effect on rankings.

4. News & daily briefs

Our daily AI news coverage is built from a curated list of primary sources: official AI lab blogs and research pages, technology news outlets with established reputations (The Verge, TechCrunch, MIT Technology Review, VentureBeat, Ars Technica), and arXiv. We scan these sources every morning and select items based on three filters:

1. Newsworthiness. Is this a capability shift, a pricing or access change, a regulatory event, or a research result that changes the state of the art?
2. Operator relevance. Can a reader do something with this? Use the new model? Adjust a strategy? Apply for access?
3. Independent verification. Can we confirm the claim against at least one primary source?

Brief-format articles run 600–1,200 words and focus on "what happened, why it matters, what to do." Deep-dive articles run 1,500–2,500 words and include independent analysis. Both formats carry the same sourcing standards.

5. Use of AI in our workflow

Because we cover AI, we also use AI. The specifics:

• Research & discovery. Large language models (primarily Anthropic's Claude) help us scan official RSS feeds, identify the day's most relevant items, and extract key facts.
• Drafting. AI tools produce first drafts of news briefs from primary sources. Every draft is assigned to a human editor for review before publication.
• Review. The human editor verifies every factual claim against the linked source, rewrites for clarity and house voice, adds any independent analysis, and signs off the byline.
• Visuals. Where we use AI-generated images, captions state "Image generated with [tool]."

What we don't do: publish unreviewed AI output, fabricate quotes or statistics, or pretend AI writing is fully human-authored. We think "AI-assisted, human-edited" is the right model for a small publication covering a fast-moving beat, and we'd rather be transparent about it than hide it.

6. Corrections policy

We make mistakes. When we do, we fix them publicly:

• Material errors (incorrect facts, wrong names or numbers, misquoted sources) are corrected inline and flagged at the bottom of the article under a "Correction" note, including the date and what was changed.
• Minor typos or formatting are fixed silently.
• Retractions. If a whole piece turns out to be based on a false premise, we will publish a retraction in place of the article, explaining what went wrong and what we're doing to prevent it.

To report an error, email corrections@aitechspectrum.com. We aim to acknowledge reader corrections within one business day and publish fixes within two.

7. Advertising & affiliate

We run Google AdSense display ads and participate in affiliate programs including CJ Affiliate, Impact, and Amazon Associates. All advertising is:

• Labeled as "Sponsored content."
• Separated from editorial — advertisers do not see or influence editorial content.
• Subject to our own advertising standards: we decline ads for adult content, weapons, gambling, unregulated crypto schemes, and other categories detailed in our Affiliate Disclaimer.

Affiliate links are marked with a disclosure box at the top of every article that contains them. Our full Affiliate Disclaimer explains which relationships are active and how commissions work.

8. Conflicts of interest

Our team members hold personal opinions and, in some cases, small positions in publicly traded tech companies. Anyone writing a review of a company in which they hold more than a nominal financial interest discloses it in the piece's byline. We do not accept equity, advisory roles, or consulting relationships with companies we cover.

9. Contact the editorial team

Corrections: corrections@aitechspectrum.com
Tips: tips@aitechspectrum.com
Editorial feedback: editor@aitechspectrum.com
Press inquiries: press@aitechspectrum.com

These standards evolve. Substantive changes will be logged in a changelog at the bottom of this page when we next revise it.