# BirdNET V3.0 Acceptable Use Policy ## Purpose BirdNET is an open scientific project developed to support biodiversity research, wildlife conservation, environmental education, and citizen science. This policy applies to the BirdNET V3.0 models — the trained model weights and any converted or quantized versions of them, such as TensorFlow/LiteRT, TorchScript, and ONNX exports. We encourage broad adoption of the BirdNET V3.0 models for lawful and responsible purposes that contribute to the understanding and protection of nature. ## Responsible Use Users should: - Comply with all applicable laws and regulations. - Respect privacy and local recording regulations, including where autonomous recording units are deployed. - Use BirdNET predictions responsibly, recognizing that model outputs are probabilistic and may contain false positives or false negatives. - Verify observations where appropriate before making scientific, conservation, or management decisions. - Consider the potential ecological impact of publishing sensitive species locations. - Account for the additional accuracy tradeoffs introduced by reduced-precision (FP16, INT8) versions of the models when interpreting results. ## Unacceptable Uses BirdNET should not be used to facilitate activities that harm wildlife, ecosystems, or people. Examples include, but are not limited to: - Assisting or facilitating poaching or the illegal capture, trade, harassment, or exploitation of wildlife. - Intentionally locating endangered or sensitive species for harmful purposes. - Supporting activities that violate conservation laws or protected-area regulations. - Using BirdNET as part of weapons systems, military targeting, or military surveillance intended to cause harm. - Using BirdNET-based devices or applications for covert surveillance of people. - Knowingly using BirdNET in ways that violate applicable laws or regulations. ## Scientific Integrity Users are encouraged to: - Cite BirdNET in scientific publications. - Report suspected model errors or biases. - Share improvements and contribute back to the project whenever possible. - Clearly distinguish model predictions from verified biological observations. - Report the model version and precision used, so results remain reproducible. ## Branding Forks, derivative applications, or redistributed versions should not use the name "BirdNET", "BirdNET+", "BirdNET V3.0", associated package names, or official project branding in a manner that could reasonably imply endorsement by the original project or its maintainers. Please choose a distinct name and branding for derivative applications unless explicit written permission has been granted. ## Feedback We welcome reports of misuse, security concerns, or ethical issues. Contact: ccb-birdnet@cornell.edu