Is it allowed to use ChatGPT or Copilot in a hackathon?

It depends entirely on the event's rulebook. Published rules range from open permission with disclosure, to debugging-only, to discouraging bulk code generation. There is no single standard. Read the rules page of your specific hackathon, and if the rulebook is silent, ask the organisers in writing before you build.

Do I need to disclose that I used AI?

In many published rulebooks, yes. MLH's example rule requires a statement in the submission. The CodeRED hackathon rules ask teams to be honest and transparent about the AI tools they used. TCS CodeVita separately requires you to declare any external code source. When in doubt, disclose what you used and for which parts.

What is the explain test?

It is a rule used by one school hackathon's AI policy: if a judge asks about any part of your code, every team member should be able to explain what it does and why it is there. If you cannot explain it, the policy says you should not submit it. It is a useful standard even when your hackathon has no such rule.

Will using AI get me disqualified?

Only if you break your event's rules. Some hackathons forbid AI for core parts of the project or restrict it to understanding errors, and some penalise or disqualify projects with excessive AI-generated code or undisclosed use. Others explicitly allow code assistants. Follow the written rules, and keep a record of how you used AI in case judges ask.

Can You Use AI Coding Tools in a Hackathon? What Real 2026 Rulebooks Say

Q: What does MLH recommend about AI use at hackathons?

MLH's guide for organisers tells them to decide their AI policy and state it clearly. Its example rule says hackers may use LLM or AI tools but must state how they did so in their Devpost submission, that a project should not be a reskin of an existing AI tool, and that unreported AI use can lead to disqualification and a report to MLH.

Whether you can use ChatGPT, GitHub Copilot or another AI coding tool in a hackathon depends on that hackathon's rulebook. There is no universal rule. Published 2026 rulebooks fall into roughly three groups: allowed with disclosure, limited to debugging and understanding, and discouraged for bulk code generation. If you are unsure, ask the organisers before you start building.

What MLH tells organisers to do

MLH, which supports many student hackathons, publishes a guide on writing hackathon rules. It lists questions organisers should settle in advance, including how many people can be on a team, whether hackers can use publicly available frameworks, and what the policy on LLMs, ChatGPT and AI is.

Its example wording says hackers may use LLM, ChatGPT or AI, but must state how they did so in their Devpost submission. It adds that a project should not be a reskin of an existing AI tool, and that if you do not credit your AI usage, you will be disqualified and reported to MLH's cheating address. That is an example for organisers, not a rule that binds every hackathon.

What real rulebooks say

Allowed, with honesty. The CodeRED hackathon rules, published on GitHub, say teams may use AI to assist while coding, including code completion, code generation and image generation, and that teams should be honest and transparent about the tools they used. The same rules require that work is done during the hacking period. A corporate hackathon's rules from Exadel, held in April 2026, permit AI code assistants such as Copilot, Codeium and ChatGPT, subject to data and security rules.

Limited to understanding and debugging. The Chakravyuha Fest 2K26 hackathon at Srinivas University restricted AI tools to error understanding and debugging. Another student hackathon's rules allowed help for brainstorming, debugging and clarifying concepts, but said that using AI to generate core parts of the project, such as major portions of code, designs or written content, was not allowed.

Discouraged for bulk generation. One school hackathon's AI policy allows learning, debugging, documentation lookup and brainstorming, but discourages generating entire components, copy-pasting without understanding and relying on AI for most of the codebase. It says projects with excessive AI-generated code may be penalised or disqualified at the judges' discretion, and proposes the "explain test": every team member should be able to explain any part of the code.

Coding contests are different again

Timed coding contests tend to care about where your code came from. TCS CodeVita's official FAQ says that if you use code from the internet you must declare the source before submitting, with precise URLs, or the code might be treated as plagiarised. It does not describe an AI policy on that page, so read the full rules before the contest.

What about SIH?

Reskilll's SIH 2026 advice tells teams that judges expect AI integration and suggests using AI coding tools to build faster in the 36-hour finale. That is a blogger's tip, not an official SIH rule, so check the SIH problem statement and finale instructions.

A quick checklist

Read the rules page for the words AI, LLM, Copilot, generated and original.
Find the disclosure rule. Some events ask for a statement in the submission.
Check the originality rule. Many events ban pre-built projects or work started before the hacking period.
Keep a log of which tools you used and for what, since judges can ask.
Be able to explain every line you submit, whatever the rules say.
Ask the organisers in writing if the rulebook does not answer the question.

Where this fits

Brabble's posts on cheating in online assessments and on proctoring software cover integrity in coding tests. For hackathons, the pattern is simpler: follow the written rules, disclose what you used and understand what you ship.