Eyn un Tsvantsik is a regular column by Sam Bolitzer on young Jewish American life in the 21st century amid larger cultural and political trends, past and present.
Artificial intelligence seems to influence every aspect of our lives in 2026. We get ads for chatbots when watching television and online videos. We see unappetizing AI-generated food on posters for local cafes. We even have AI shoved in our faces when trying to work; as I write this very article in Google Docs, I see a button encouraging me to use Google’s LLM Gemini in the upper right corner. This omnipresence is even more noticeable to us as college students. Although many courses have policies prohibiting AI, with so many of our classmates using the technology for assignments, it can sometimes feel like we need to use AI in order to keep up. This trend is certainly encouraged by the Trump administration, which declares concern about AI to be a “hoax” aimed at hindering American competition with China.
The moral dilemmas that AI use presents have been widely discussed in the past few years. However, it may be worthwhile, at least for myself and my fellow Jews on campus, to reflect on the religious implications of AI use. As we are thrust into an era where AI is presented to us as inevitable and necessary to succeed, we should remind ourselves that prior millennia have plenty to say about these new tools.
Judaism is blunt in its condemnation of plagiarism as a form of theft; specifically, Halakha tells us that plagiarism falls under the category of geneivat da’at, which literally means “theft of the mind.” With this idea, there seems to already be conflict between Judaism and AI; LLMs’ habits of plagiarizing have been widely written about, as the models’ outputs are formed (or at times outright copied) from potentially millions of materials included in its training data, with often no credit given to the contributing texts in those responses. What could be a more quintessential example of geneivat da’at than an algorithm literally designed to plagiarize?
But is this too harsh? After all, you can ask an LLM to provide citations for every piece of information that it gives you, and disclose that you used it in your assignments—how could you be plagiarizing then? But you can never be truly sure which materials an AI is pulling from, for both the information that it provides and the very structure of its responses, especially as many of those materials are physically destroyed in the process. I was reminded of this when I came across this text in the Mishnah, specifically Chullin 89a:4:
“Difficult is the return of theft that has been consumed… it is stated: ‘That I will not take a thread or a shoe strap nor anything that is yours… except only that which the young men have eaten with me’ (Genesis 14:23-24). Even the righteous Abraham was unable to return that which the young men had already consumed.”
Indeed, AI ingests data like we eat food. If you find out that food you’ve eaten was stolen, you can’t exactly “return” it from your stomach. Likewise, once a piece of media has been added to an AI’s training data, it can’t be “returned”; it is now influencing that AI’s outputs, down to its fundamental writing choices. Can we expect ourselves to cite every single piece of training data that could have influenced or contributed to an LLM’s response that we use in an assignment? Certainly not.
But maybe using AI just a couple of times for an assignment is not that bad. There is a difference between including a few AI-edited sentences in your term paper and generating the whole thing with a chatbot. How harmful can the former really be? To answer this question, we can turn to Taanit 16a:13, a passage in the Talmud discussing the Book of Jonah:
“Shmuel said that the king of Nineveh proclaimed: Even if one stole a beam and built it into his building, he must tear down the entire building and return the beam to its owner… these people wanted to repent completely by removing any remnant of stolen property from their possession.”
We can easily compare this image to that of writing a paper. Each sentence and idea is like a beam of a building, supporting the wider structure of a text; if a single one of them came not from the writer’s own mind or their well-sourced research but from an LLM, then the integrity of the whole text is in question. The impossibility of fully “returning” the theft inherent in AI means that it would be best to not include text generated, edited, or otherwise influenced by AI in assignments at all.
AI is here to stay. The biggest proponents of this new technology are often quick to brand its critics “luddites,” arguing that we should get used to AI’s new power over society and simply accept it as the new normal. But we have a responsibility to reject practices which are unethical, regardless of their potential advantages. As Jews, religious or not, we have often looked to our legal and cultural traditions to help us navigate new developments in wider society, allowing us to refrain from embracing troubling practices that others might. When we seek to emulate such harmful practices to “get ahead” regardless, it can be disastrous; it’s no surprise that Israel, having similarly adopted external frameworks of ultranationalism, has utilized AI as a tool in the Gaza Genocide.
I will admit, in the past I have occasionally used AI for things that I should not have. It became a tempting tool as I drowned in depression. Yet as I rebuild myself in my junior year and take back control of my life, I have looked to my cultural and intellectual ancestors for guidance. A central piece of advice I have found in them is to refrain from AI use. I would encourage all of us, Jewish or otherwise, to do the same.
