Similar to a bug bounty program, we are launching a false positive and AI detector bypass bounty program.
Leading AI detectors are highly accurate but not perfect. Our latest models are highly accurate, and instances where our model fails are harder to find.
We have received useful edge cases from customers in the past that have improved our AI detection. For example, early on, Unicode characters were able to trick our detector (no longer), and Markdown files reduced the detector’s accuracy. Both of these cases were shared by users and helped us improve later versions of our AI detection models. This program is meant to reward users for helping us understand the edge cases where our tool is not perfect.
Submit verifiably human-written text that Originality.ai incorrectly identifies as AI (a false positive). Approved submissions receive $100 in Originality.ai credits.
Submit False Positive Bounty Here
Submit a method that consistently bypasses our AI detector for consideration for a monetary reward.
These reports need to include exact steps so our team can reproduce the AI detection bypass method. A systematic, high-value finding may receive a monetary reward at our discretion.
As accuracy improves, the remaining repeatable failures become both harder to find and more valuable. A good report gives our AI research team a new case to investigate.
To submit a finding, use the program’s Google Form.

We tested 1,000 English GPT-6 Astra responses with Originality.ai AI detector to find out if GPT-6 Astra content was detectable. We also compared length and readability with Kimi K3 and archived responses from OpenAI, Anthropic, DeepSeek, Google and xAI on the same prompts.