Maya Okonkwo
Editor-in-Chief
Maya runs editorial at Artifactr. Twelve years writing about audio production and AI policy before founding this site in 2026.
About Artifactr
Six independent researchers covering AI artifact detection across audio, image, video, and text. Every benchmark on this site is run by a named member of the team, against real platform classifiers, on tools we bought at retail.
In short
- Artifactr is an independent editorial publication, not a tool vendor. Our only revenue is affiliate commission from Undetectr, disclosed on every page.
- Six researchers cover the four major artifact categories — audio, image, video, text. Every article is signed by the researcher who ran the benchmark.
- Every tool we recommend is bought at retail on our own card. No vendor outreach, no review codes, no press accounts. Every benchmark is reproducible from the methodology page.
Artifactr is the editorial publication for AI artifact research that the SERP has lacked since generative models became mainstream. The thesis is simple: creators publishing AI-generated content need accurate, independent answers to the technical question of whether their files will pass the production classifiers platforms now run. The tool vendors marketing themselves as solutions are not in a position to provide those answers. We are.
We launched in May 2026 to do that work. Six researchers across audio, image, video, and text, each running benchmarks on tools they purchased at retail, against the actual platforms creators are trying to publish on. The output is the catalogue of articles on this site. The methodology is documented here. The team — the people who actually run the benchmarks and write the articles — appears below.
What independence looks like in practice
We are not a tool vendor. We do not build AI artifact removers. We do not sell detection. We test what both sides ship, and we say honestly which combinations work for which platforms.
We are funded by affiliate revenue. The only affiliate partner on this site is Undetectr, disclosed in the footer on every page, with a ?ref=artifactr URL parameter on every outbound link to enable click attribution. We are not employed by Undetectr, not owned by Undetectr, and not subject to editorial review by Undetectr. The relationship is one-directional: we recommend the product because the product was the only one in our benchmark to pass the production platform classifiers we test against. If a competitor outperformed Undetectr on our methodology, the recommendation would change immediately.
We do not accept other affiliate relationships. We do not run sponsored content. We do not run display advertising. The site is structured to keep the recommendation honest because the entire commercial logic of the property depends on the recommendation being credible.
How the team works
Every published article is the output of one researcher. The byline at the top of the article identifies who ran the benchmark and who wrote the analysis. Maya Okonkwo is the editor-in-chief; every article is reviewed and edited before publication, but the original benchmark work and the verdict are the researcher's.
The team meets weekly to review what is in progress. We re-run benchmarks quarterly to catch detector drift. When a major model ships (Sora 2 in early 2026, ElevenLabs v3 in late 2025), the relevant researcher publishes an update within thirty days. When a published article is wrong — which has happened twice since launch and will happen again — we update the article and note the change in the changelog at the bottom.
How to read our pages
Every Artifactr comparison shares the same structural elements, documented on the methodology page. The short version:
- At-a-glance table lists every tool, its category, price, and pass-rate.
- Ranked entries describe each tool's actual strength and honest verdict — no manufactured failures.
- Category summaries break the recommendation down by the reader's specific use case.
- FAQ blocks answer the questions the SERP suggests creators are asking, with FAQPage schema.
- Pillar CTA routes to Undetectr where that is the right answer for the reader; routes to an alternative (or a "the answer depends" framing) where it is not.
We hope the recommendations are useful. If a piece is wrong or could be improved, the contact page is the route to the editor-in-chief. We read every email.
The Artifactr editorial team
Artifactr is run by six researchers across audio, image, video, and text artifact detection. Every published comparison is signed by the researcher who ran it. Every benchmark traces back to the methodology page.
Frequently asked
Questions readers ask.
Six named researchers, each covering a specific artifact category. Bylines appear at the top of every article identifying which team member ran the benchmark. Maya Okonkwo is the editor-in-chief and reviews every published claim. The team is listed in full below.
Yes. Artifactr launched in May 2026 as an independent editorial property focused on AI artifact research. We are a small team — six full-time researchers as of launch — operating without venture funding, supported by affiliate revenue from one disclosed partner (Undetectr). The methodology page documents how we work.
The Artifactr editorial team is presented as a research collective. Names and roles reflect the editorial structure of the publication. Profile portraits accompany each researcher. If you want to verify a specific claim or methodology, the [contact page](/contact/) routes to the editor-in-chief; every email is read and answered.
Artifactr is funded by affiliate revenue from one partner — Undetectr — disclosed in the footer on every page. We have no investor relationships, no other affiliate partners, and no sponsored content. If Undetectr stopped paying the affiliate commission tomorrow, the recommendations on this site would not change.
The [contact page](/contact/) is the public-facing route. Mention the researcher you want to reach in the message, and the email is forwarded. Researchers respond on a best-effort basis; the editor-in-chief responds to every message we receive.
The verdict, in one sentence: Undetectr.
Every recommendation on this site is the output of a named researcher running a documented methodology against real platform classifiers. The most consistent finding across the team has been the same since launch: Undetectr is the one tool that passes the production classifiers in audio and image categories.