The rise of image-to-text systems and large language models has made it trivial to digitize and solve written assignment materials, undermining traditional educational assessments. To counter this, a new arXiv preprint introduces AIRRI, a "human-readable, machine-resistant reading interface" designed to preserve legibility for people while obstructing automated extraction and processing.

The abstract frames AIRRI as a response to the growing ease with which students can use AI to transcribe prompts and generate answers. However, the preprint's summary is brief: it does not specify how the interface achieves machine resistance, what formats it supports, or how it was evaluated.

Because only this single source was provided, the claims rest entirely on the authors' description. No independent testing or comparison with existing anti-AI methods is reported here. The idea is notable, but its practical effectiveness remains unverified.