Frequently Asked Questions

Made By Humans is a professional content verification platform for publishers, journalists, and editorial teams. It detects likely AI-generated text, AI-generated or manipulated images, deepfake video, and synthetic audio — giving editorial teams confidence in the authenticity of content before publication.

The platform is built for newsrooms, publishers, editorial teams, and any organisation that needs to verify content authenticity. It's used by editors reviewing freelance submissions, fact-checkers validating source material, and compliance teams maintaining audit trails.

Detection accuracy varies by content type and method. Our image detection service (IdentifAI) reports 99.7% accuracy on their test sets. Text detection uses Pangram's models which achieve high precision on longer texts. Results are probabilistic — they support editorial judgement rather than replacing it. We never claim 100% certainty.

Uploaded images are stored securely in Google Cloud Storage for your access and audit trail. Text is stored as part of your submission history. All data belongs to your account and is not shared with third parties. You can export or delete your data at any time.

All AI detection tools have false positives and false negatives. Made By Humans is designed to flag risk and support editorial decisions — not to make them for you. We recommend using results alongside your editorial judgement, especially for borderline scores.

The platform detects images from all major generators including DALL-E, Midjourney, Stable Diffusion, Adobe Firefly, and others. It also detects manipulated photographs — images that started as real photos but have been altered using AI tools.

The manipulation heatmap highlights specific regions of an image that show signs of AI generation or tampering. Red/warm areas indicate higher suspicion. This helps editors understand exactly which parts of an image may have been altered.

Text analysis examines linguistic patterns, structural features, and statistical signals across your text. It works at the sentence level, highlighting which sections show AI-generation patterns. The overall score represents the probability that the text was machine-generated.

Yes. The authorship fingerprinting feature analyses style consistency throughout a piece. If sections were written by AI while others are human-authored, it will flag mixed authorship. The segment-level scores in text detection also reveal which specific sections score highest for AI patterns.

Video detection supports most common formats accessible via URL (MP4, WebM, MOV). Audio verification supports MP3, WAV, M4A, and OGG. Files must be accessible via a public URL for processing.

The Editorial Review tool is an AI-powered forensic reviewer. It reads a submitted article and returns a structured brief identifying unverified claims, unsourced statistics, logical inconsistencies, and passages that read as AI boilerplate. It gives editors a pre-publication checklist — it doesn't rewrite or judge the article.

Authorship analysis examines writing style metrics (vocabulary diversity, sentence patterns, transition word usage, paragraph structure) to estimate which AI model likely generated the text. It can distinguish between ChatGPT, Claude, Gemini patterns and detect mixed human/AI authorship.

Every analysis has a fixed token cost. You buy tokens in packs and spend them as you use the platform. Text analysis costs 2 tokens per check. Image detection costs 2 tokens. Heatmaps cost 5 tokens. There are no subscriptions or hidden fees — you only pay for what you use.

Starter: €2 for 20 tokens (€0.10/token). Professional: €20 for 500 tokens (€0.04/token). Newsroom: €100 for 3,500 tokens (€0.029/token). Volume discounts mean higher plans are significantly cheaper per analysis.

No. Tokens never expire. Once purchased, they remain in your account until used.

Yes. Publisher organisations have a shared token balance that all team members draw from. The publisher admin can monitor usage per member and set up auto-top-up rules.

The platform checks your balance before starting any analysis. If you don't have enough tokens, you'll be told the cost and your current balance. No analysis runs without sufficient tokens available.

The WordPress plugin adds a verification panel to your post editor. You can check article text and images directly from WordPress before publishing. It can automatically add a 'Verified by Made By Humans' badge to published articles that pass verification. Configuration is done via the plugin settings panel.

Yes. The REST API supports any CMS or editorial workflow. Authenticate with an API key, submit content for analysis, and receive results via JSON. The API supports all platform features — everything you can do in the web interface is available programmatically.

The trust widget is a lightweight JavaScript snippet you add to your website. It displays either a publisher certification badge (showing your overall trust level) or per-article verification badges. It works on any website — WordPress, custom CMS, static sites, etc. It's complementary to the WordPress plugin badge.

Watchdog lets you add RSS feeds to monitor continuously. When you trigger a scan, it fetches the latest articles from each feed and runs AI detection on them. Use it to monitor your own publication, competitor feeds, or specific contributor bylines.

Submit up to 50 URLs or a sitemap URL. The platform extracts article text from each page and runs AI detection. You get a summary showing which articles are flagged. Ideal for auditing your publication archive or checking an entire site.

Yes. The EU AI Act requires disclosure of AI-generated content. Made By Humans provides a complete audit trail showing what was checked, when, by whom, and what the results were. The compliance report endpoint generates structured data suitable for regulatory submissions.

Compliance reports cover any date range and include: total submission counts, breakdown by type (text/image), detection results, verification badges issued, team member activity, and exportable CSV data. All timestamped with user attribution.

Publishers automatically earn certification levels (Bronze → Silver → Gold → Platinum) based on their verification activity. Higher levels indicate more consistent, comprehensive content checking. The public certification page and embeddable badge let you communicate your commitment to content authenticity to readers and advertisers.