I have a new number theory formula. I am a software engineer, but I am interested in number theory. I do not have an endorsement on arXiv, so I am left with two popular choices, OSF and preprints.org. Which one is safe, so my formula will not get stolen?
I fear you have a misconception about intellectual property and publishing. The most important thing to keep in mind is that you don't and can't "own" ideas. Publishing (somewhere that has a firm date attached) can associate your name with an idea and give you a claim to priority, which is really a social construct. It is proof against plagiarism but not against use and extension.
Publishing an idea, anywhere, doesn't give you exclusive rights to it. Any one can use it. Anyone can extend it. Anyone can dispute it. Ideas are free.
There are certain protections that can be applied to certain expressions of ideas. This permits the originator to exploit the ideas financially for a time. Copyright protects the specific expression of an idea (words, figures, ...) so that creators can be rewarded for their ideas, even though they don't own those ideas.
Patent law protects certain expressions of ideas ("devices") for a time, again to permit commercial exploitation.
But, if someone needs to hold ideas "close" so that others can't take any advantage of them at all, they are held as trade secrets and not published. Some (likely weaker) protections apply to these but in the form of sanctions against those who reveal them without permission.
The very purpose of publishing is to make the ideas public and available to others for use and extension.
Publishing anywhere in a venue that provides a verifiable date of publication and is stable (i.e. not a personal website) gives you a "receipt" that can be proof against plagiarism of others. But plagiarism isn't the use of the ideas, nor their extension, but only an improper claim of priority.
And, keeping the ideas to yourself, while it may prevent others from knowing what you know, has the disadvantage that your discovery might be published through the independent work of others. You then have no proof of priority.