Pointing a math app at your homework is not cheating by default. It becomes cheating the moment you use it on work your teacher said you had to do alone, which is a rule that exists in your course syllabus and has nothing to do with the app itself.
Nobody likes that answer. Students were hoping for permission. Teachers were hoping for a ban. What they get instead is a rule that is narrower than either side assumes, and once you can see where it sits, a different question starts to matter a lot more than this one.
The short answer, in one rule
Ask one thing: for this specific assignment, did the instructor authorize outside help?
If the answer is yes, or if the assignment is ungraded practice, using a solver is no different from checking the back of the textbook. If the answer is no, or the assignment is a test, quiz, or anything labeled independent work, then using a solver is unauthorized assistance and you are over the line. Not because it is an app. Because you were told to work alone and you did not.
The confusing part is that almost nobody states this rule out loud. Syllabi say "no outside help" and assume everyone agrees on what help means. Students hear "no copying" and assume a step-by-step explanation is something else. Both readings are reasonable, and they do not match.
What academic integrity policies actually say
Read the policy at almost any university and you will notice something: the apps are not named.
The University at Buffalo, for example, defines its academic integrity violations around behavior, not products. Cheating covers using unapproved materials during an assessment, and the examples given are ordinary objects: calculators, smart watches, phones, other students' work. The one technology written into the policy by name is artificial intelligence, and even that appears as a clause about submitting an assessment "prepared by any person or technology" other than the student.
This is deliberate. A policy that listed apps would be out of date within a semester. So the language stays general, which produces a rule you can actually apply: the question is never "is this tool banned," it is "was I permitted to get help here."
Two consequences follow, and students consistently get both wrong.
First, a tool being free, popular, or recommended elsewhere gives you nothing. Your history professor allowing a grammar checker says nothing about your calculus homework. Permission is per assignment, per instructor.
Second, "I only used it to check my answer" is a real defense in some courses and no defense at all in others. If the assignment is graded on process, checking the answer changes the work you submit. Ask, in writing, before the deadline rather than after.
The question that matters more
Most students searching whether Photomath counts as cheating are not really asking about ethics. They are asking whether they will get caught. That is the wrong worry, because the more likely cost shows up on a test where nobody accuses you of anything.
Here is the trap. Reading a correct solution feels exactly like learning. Every step follows the one before it, nothing snags, and you put the phone down honestly convinced you have got it. Psychologists call that feeling fluency. It tracks how easily something went down, not whether you could produce it yourself on a blank page next month.
The evidence for this is twenty years old and has not budged. In their 2006 work on test-enhanced learning, Henry Roediger and Jeffrey Karpicke put one group of students through repeated restudying and another through repeated self-testing. Test them five minutes later and restudying looks better. Wait a while, and the result flips hard: the students who practised pulling the material back out of their own heads retained substantially more, on less time with it.
Watching a solver work through your homework is the restudying condition. It is the version that looks better on Tuesday night and worse in November.
So the practical risk is not a misconduct hearing. It is arriving at an exam with twenty problems you have seen solved and zero you have solved.
Four situations, four different answers
A graded problem set where the syllabus says work independently. Off limits. This is the clear case and there is no clever reading of it.
A test, quiz, or proctored exam. Off limits, and this is the version that ends careers rather than grades. Do not bring the phone.
Practice problems you assigned yourself, or homework marked for completion rather than correctness. Fine. This is closer to using an answer key, which teachers have handed out for a century.
Checking your own finished work. Usually fine, occasionally not, and worth one email to find out. In my experience teaching secondary math, this is the use that separates students who improve from students who plateau, and it is the one they most often assume is forbidden.
Notice that in three of the four cases, the app is not the variable. The assignment is.
How to use a solver without hollowing out your skills
If a solver is permitted, the order you do things in decides whether it helps you or replaces you. The sequence that works:
- Attempt first, on paper, badly. Even a wrong attempt marks where your understanding stops, which is the only information the solution can attach to.
- Solve the problem, then close it. Read the steps once for the method, not for the answer to copy down.
- Redo it blind. Blank page, no reference. This is the step almost everyone skips, and it is the entire difference between the two conditions in the retrieval research above.
- If the blind redo fails, you did not learn it. That is useful, not shameful. Go back to the step where you stalled and ask why that step is allowed rather than what it was.
- Then do a fresh problem of the same type. One you have never seen worked. If you can do that one, you own the method.
The ask-why part is where a chat-based tool beats a static answer key, because you can push on a single line. Working through how to isolate a variable step by step and then asking why a particular move was legal gets you closer to the exam than any amount of rereading. The same applies across topics: our step-by-step topic calculators exist to explain a method, not to hand you a number to transcribe.
One practical thing about Photomath, since that is what most people mean by "the math app." It lives on your phone and nowhere else. Google finished buying it back in 2023, and even now the site just points you at the App Store or Google Play. There is no browser version to open. The step-by-step explanations do not cost you anything. Which means that on a laptop, you are either reaching for your phone or you are using something else entirely.
If you change one thing
Ask your instructor, once, in writing: "For this course, am I allowed to use a step-by-step solver to check my work after I have finished it?" You will get a yes or a no, the ambiguity disappears for the whole semester, and you have a record either way.
Then add the blind redo to your routine. Not because it is virtuous, but because the alternative is a study session that feels productive and leaves nothing behind. The students who use these tools well and the students who get wrecked by them are often using the same app for the same amount of time. The difference is whether the page was blank the second time. If you want to practice that on messier material, translating a word problem into equations is where the gap between recognizing a method and running it shows up fastest.