[Biopython] Generative AI policy for contributions to Biopython
Andrew Dalke
dalke at dalkescientific.com
Fri Apr 24 08:53:15 EDT 2026
On Apr 24, 2026, at 13:43, Markus Piotrowski <Markus.Piotrowski at ruhr-uni-bochum.de> wrote:
> I wonder how you really can prevent this.
The same way Biopython prevents someone from contributing code without permission of the copyright holder, or code which implements an algorithm under patent protection - say "no", be aware that people may do it, and if that happens and is found out, do what can be done to remove it.
> Am 24.04.2026 um 12:16 schrieb Peter Cock:
>> I blogged about my views late last year - ending in the line "Right now, I
>> still lean very much to saying no any PR using generative AI".
I'm one of those contributors, leaders, and maintainers who have gone and went. I know I have no say at all about the project direction.
Given what little it's worth, I support the no-AI position.
I long ago moved over to cheminformatics. I've had the displeasure of seeing AI-generated code bases in my field. One was clearly using AI for not-actually "clean room" rewrite of an existing code base, and falsely believed that doing so removed copyright protection. Another contained code obviously derived from an open source code base, but without the required attribution. "Obvious" to someone like me, who has looked at the original source and alternative implementations, but the novice developer using AI tooling likely didn't realize it was plagiarized.
Saying "the human submitter is responsible" assumes the contributor understands what it means to be responsible. As a general rule, scientists are not trained in the nuances of copyright law, nor have the time to evaluate the severe negative effects of these code generation tools while bombarded by "use AI!" boosterism, all under pressure of writing a thesis or paper.
Some 20-odd years ago, when I helped manage the BOSC talk submission process, I would get submissions saying something like "code available for academic use only". I explained why that wasn't open source, and most (all?) changed their licenses to an actual open source license. They simply weren't aware, and needed guidance.
Saying "No generative AI", and following up on it, is also a way to make people aware, and provide guidance.
Andrew
dalke at dalkescientific.com
More information about the Biopython
mailing list