AI Cheating Prevention During Contests

Hello everyone,

With the rapid advancement of AI tools, maintaining fair competition has become increasingly challenging. While no technical solution can completely eliminate cheating, I believe there are several measures that could significantly increase the effort required to misuse AI during live contests.

I would like to propose adding an optional "contest protection mode" for the duration of official Codeforces rounds.

Proposed features

1. Disable text selection and copying on problem statements

Prevent selecting text with the mouse.
Disable common copy shortcuts such as Ctrl+C.
Disable the context menu (right-click) on problem statements.

These techniques are not foolproof, but they discourage casual copying and make it less convenient to paste the statement directly into an AI assistant.

2. Make screenshots more difficult where possible Modern web browsers cannot reliably prevent operating-system screenshots, so I understand that this cannot be completely solved on the web. However, if there are browser features or experimental approaches that can make automated capture more difficult, they could be worth investigating.

Why this is useful

These protections will not stop a determined cheater. Someone can always:

Type the statement manually.
Take a photo with another device.
Use OCR or other workarounds.

However, security is often about increasing the cost of abuse rather than making abuse impossible.

Today, many AI-assisted cheating attempts are extremely easy:

Copy the statement.
Paste it into an AI model.
Receive a solution within seconds.

Even adding a small amount of friction may discourage opportunistic cheating and help preserve the spirit of the contest.

Important considerations

These protections should only be active while the contest is running. After the contest ends, the problem statement should return to its normal behavior so that participants can:

Copy sample tests.
Quote parts of the statement in editorials or discussions.
Translate or study the problem.
Use accessibility tools as usual.

Final thoughts

I fully understand that this proposal is not a complete anti-cheating solution. Client-side restrictions can always be bypassed by determined users.

Nevertheless, I believe that making AI-assisted cheating less convenient could improve contest integrity with relatively little impact on honest participants, especially if these measures are enabled only during active contests.

MikeMirzayanov, thank you for creating and maintaining Codeforces. I would like to hear your thoughts on whether this proposal could be considered.

Thank you for reading.