← Back to LitMark
Is Using AI Tools to Find Sources Academic Misconduct?
A practical guide for students · Updated 22 October 2026
Using an AI tool to find a starting point for your research isn't, by
itself, academic misconduct at most universities. Most policies target
two specific things: submitting AI-generated text as your own work, and
citing sources that don't check out. The second one is where students
actually get caught out, and it has nothing to do with whether AI was
involved at all.
The problem is that AI chatbots don't search academic databases. They
predict plausible-sounding text, which means they can generate a citation
that reads perfectly, author, journal, year, DOI, and doesn't exist, or
exists but says something different to what you've written. A 2023 study
in Scientific Reports checked 636 citations generated by
ChatGPT and found 55% of GPT-3.5's citations were entirely fabricated,
and even GPT-4's figure was still 18%, with a further quarter of its
real citations containing substantive errors. If a fabricated or
misquoted source ends up in your reference list, it's no longer an
AI-use problem. It's fabrication or misrepresentation of a source, which
is exactly what most academic integrity policies are written to catch,
regardless of how the citation got there.
Where the actual line sits
-
Using AI to find a topic, structure an argument, or point you toward
search terms is treated by most institutions the same as using a tutor
or a study guide: a starting point, not a problem, as long as the final
work and the sources in it are yours to stand behind.
-
Citing a source you haven't actually opened is the misconduct risk,
whether an AI tool gave it to you or a classmate did. If you can't find
the paper, you can't cite it.
-
A citation that exists but is misquoted is just as serious as one
that's invented outright. AI tools will confidently summarise a paper's
conclusion in a way the actual abstract doesn't support, for example
citing a small pilot study as if it proved a general effect, or
attributing a finding to the wrong population entirely.
-
Submitting AI-written sentences as your own analysis is a separate,
usually more serious breach, and most detection tools now focus on
wording patterns rather than on which tool you used to find a source.
It's also worth knowing the stakes don't reset after a first flagged
issue. A single confirmed breach doesn't just affect that one
submission, it usually means every piece of work you hand in afterwards
gets checked more closely for the rest of your degree. It's a
reputational shift, not just a one-off penalty.
How to make sure you're covered
-
Open every source before it goes in your reference list. Not the AI's
summary of it, the actual paper, abstract, or report.
-
Check the claim against the source, not just the existence of the
source. A real DOI doesn't mean the paper says what you think it says.
-
If you can't locate a paper on Google Scholar, PubMed, or your library
database within a couple of minutes, treat it as unverified and don't
cite it until you can.
Checking every source and citation by hand takes far longer than it
sounds, and doing it manually the night before a deadline tends to add
stress you don't need on top of the submission itself. It's also the
reason LitMark is built the way it is: it never generates a word of your
essay, it finds genuine papers from real academic databases instead of
inventing them, and it checks every citation against the actual source
before you submit, so what ends up in your reference list has already
been validated rather than something you're trusting blind.
The honest takeaway: the tool you used to find a source isn't what gets
flagged. The source you didn't check is. Use AI tools to get started if
that's useful, but verify everything that makes it into your reference
list against the real paper, every time.
← Back to LitMark