AI explainers
AI explainers — how generation works and where detection stands.
Plain-language background on how generative models produce images, video, voice, and music — and what their architecture implies for what detection can and cannot do. Read these once and the practical guides make more sense.
- What is a deepfake?
Plain-English explanation of deepfakes — how they're made, what they can fake, and what current detection can and can't do.
5 min read - How AI image generation works
A non-technical explanation of how diffusion models like Stable Diffusion and Midjourney create images — and what that tells us about detection.
5 min read - Why AI detectors are not 100% accurate
What every AI-detection tool can and cannot do, why false positives happen, and how to use detection responsibly.
5 min read - How accurate are AI image detectors?
What current AI image detectors get right, where they fail, and how to think about an accuracy figure when one is quoted.
5 min read - How accurate are deepfake detectors?
Deepfake-detection accuracy in 2026 — what it means, where it breaks, and why benchmark numbers consistently overstate real-world performance.
5 min read - How accurate are AI text detectors?
AI text detectors have a high-profile accuracy problem — false positives on formal English, false negatives on humanized output. Here's the realistic picture.
5 min read - Can AI detectors be fooled?
Yes — and understanding how is essential to using detection responsibly. The four main evasion patterns, and what they imply for how much weight to put on a clean score.
5 min read - Does Turnitin detect AI? Honest answer for 2026.
Yes, Turnitin has an AI detector — and yes, it has the same accuracy limits as every other AI text detector. Here's what it actually catches, what it misses, and what to do with the result.
6 min read - What is an AI humanizer, and does it actually work?
AI humanizers rewrite ChatGPT output to evade AI detectors. Here's how they actually work, why detectors keep catching up, and what 'undetectable AI' really means.
6 min read - Are AI humanizers really undetectable?
Tools promising 'undetectable AI' have a credibility problem. Here's what humanizers actually achieve against current AI detectors — and what they fail to hide.
5 min read - GPTZero vs other AI detectors: what actually works?
How GPTZero compares to Turnitin's AI checker, Originality.ai, Winston AI, and free standalone detectors. Where each one fits and where each one fails.
6 min read