Detection for the post-AI internet.

What we do

We detect AI-generated content across four modalities - text, images, audio and video - in one platform with one credit pool. Every scan uses the same six-engine cross-model ensemble whether you're checking a 30,000-word essay or a 60-second Veo 3.1 clip.

Why this exists

The detector landscape in 2026 is full of single-modality tools and paywalled trials. Originality.ai, GPTZero, ZeroGPT and Copyleaks are text-only. Reality Defender is enterprise-only. Nothing exists for the writer, editor or recruiter who needs to verify all four modalities in one workflow without committing to a $14.95 minimum.

That's the gap we filled. $1 in free scan credit on signup, the same credit pool on every model, lifetime scan history that never expires, and a feature matrix that's easy to compare against any competitor.

How we measure accuracy

Our published 99.1% accuracy and 1.2% false-positive rate come from the public RAID benchmark across GPT-5.5 and Claude Opus 4.7 generations. We re-run the benchmark monthly as new models ship and publish the deltas. The detection methodology is a cross-model ensemble combining tokenisation patterns, perplexity, burstiness and a learned classifier.

What we don't do

We don't sell humanisers. We don't sell evasion. We don't train on your data. We don't have a retention tier - your scan history is forever on every plan, including Free. If a feature isn't on this site, we haven't shipped it.

The team

Built by a small team of engineers and applied researchers distributed across Europe and North America. Headquartered in Delaware (USA). We respond to email - hello@ai-detectors.io - usually the same day.

What's next

More models as they ship - Midjourney v8 Alpha (March 2026 launch), new Claude and GPT versions, Veo and Kling updates. Browser extension, teams and SSO are explicitly out of scope right now; we'd rather do four modalities exceptionally well than fifteen features mediocrely.