OpenAI Decompiling Copyrighted Code

3 points by bubblebobble 2 years ago | 3 comments

Hello everyone. I recently received a notification from OpenAI indicating that one of my API keys was found on GitHub and has been deactivated.

Shutting down the APIKEY itself was not the issue (let's face it, mistake on our side).

The issue arises because the API key was contained within a copyrighted Java jar file, which was hosted on a public repository and used as a dependency. This situation suggests that OpenAI is decompiling packaged code, including jar files, and potentially reverse-engineering minified JavaScript files that are copyrighted.

In essence, I'm worried that OpenAI may be accessing and utilizing copyrighted code for training purposes. (why else would they be decompiling other peoples code? I can't see they would implement decompilation and throw away an opportunity)

Any thoughts? Are there a recommended approach or any channels through which this issue can be addressed?

TheJCDenton 2 years ago |

Finding printable characters in some archives or even machine code can be done without decompiling anything. Unix tools like strings[1] are usually used for this kind of work.

[1] https://en.wikipedia.org/wiki/Strings_(Unix)

bunbun69 2 years ago |

> it seems that OpenAI may be accessing and utilizing copyrighted code for training purposes.

There's currently no proof that they do. They only decompiled your app to see if you misuse the API key... which you do. You shouldn't ship API keys

bubblebobble 2 years ago | |

Thanks Ill update my wording. It seems funny that OpenAI would decompile jar files to source code for simple fun and checking API keys. The primary AI use with github of course is code training data....