# I am concerned about LLM code in Python

**URL:** <https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691>\
**Category:** Python Help\
**Created:** [March 26, 2026, 8:45am UTC](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691 "2026-03-26T08:45:20Z")\
**Posts on this page:** 4\
**Page:** 8

<div class="post-metadata">

**Author:** ![tim.one](https://sea2.discourse-cdn.com/flex002/user_avatar/discuss.python.org/tim.one/32/282_2.png) [@tim.one](https://discuss.python.org/u/tim.one)\
**Post date:** [April 5, 2026, 4:00am UTC](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691/153 "2026-04-05T04:00:58Z")

</div>

LOL! I just got an ad on Facebook for “Rewritify”, an online business that, for a price, offers to rewrite your AI output in ways that evade AI-detection technologies. Yet another arms race 😉

I pointed some “free”’ detectors at parts of CPython’s code base, but no joy. They all said too much text for the free version, but would I like to sign up for a monthly subscription? Not really, no. One offered to check `unicodedata.c` for a one-time US$2.80. I was tempted, but passed.

Why that? Because Unicode is a very involved standard, and implementations attract few eyeballs. How could you write this stuff _without_ largely duplicating many other implementations of, say, Hangul syllables?

My understanding, though, is that it probably doesn’t matter - that courts already recognize that copyright doesn’t apply in technical areas with very constrained choices for how a thing can be expressed. For which reason you should stick to variable names like “i” and “j”, not like “hangul\_syllable\_index\_1” - the latter may be viewed as protected “creative expression”‘’ 🤣

---

<div class="post-metadata">

**Author:** ![Lucas\_Malor](https://sea2.discourse-cdn.com/flex002/user_avatar/discuss.python.org/lucas_malor/32/24845_2.png) [@Lucas\_Malor](https://discuss.python.org/u/Lucas_Malor)\
**Post date:** [May 29, 2026, 9:05pm UTC](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691/154 "2026-05-29T21:05:31Z")

</div>

It seems that exist some open source LLMs, like Llama. But who’s is interested in train such a model for detecting other AIs code?

Anyway, I don’t know a programmer that didn’t copy-pasted some piece of code from somewhere on internet. How can you be sure that code is not copy-pasted from some other source under a license?

About the fact the AIs don’t create but just copy, this os what they want you believe 😉

Furthermore, also humans copy a lot. It’s the better way to learn. God bless the mirror neurons.

---

<div class="post-metadata">

**Author:** ![adrianofdoom](https://sea2.discourse-cdn.com/flex002/user_avatar/discuss.python.org/adrianofdoom/32/22114_2.png) [@adrianofdoom](https://discuss.python.org/u/adrianofdoom)\
**Post date:** [June 30, 2026, 11:57am UTC](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691/155 "2026-06-30T11:57:31Z")

</div>

> [@sirosen](#):
>
> Bans which are unenforceable are a tactical mistake. You guarantee that the rule will be broken silently, behind your back.

Absolutely.

---

<div class="post-metadata">

**Author:** ![Dutcho](https://sea2.discourse-cdn.com/flex002/user_avatar/discuss.python.org/dutcho/32/7074_2.png) [@Dutcho](https://discuss.python.org/u/Dutcho)\
**Post date:** [June 30, 2026, 10:04pm UTC](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691/156 "2026-06-30T22:04:01Z")

</div>

Also see [Should Claude Code's usage be described in the code/docs somewhere?](https://discuss.python.org/t/should-claude-codes-usage-be-described-in-the-code-docs-somewhere/107969)

[Previous page](https://discuss.python.org/t/i-am-concerned-about-llm-code-in-python/106691.md?page=7)