Get Started
Home
Topics
Search
Library
Research questionHow can rubric-based LLM grading resist prompt injection without compromising fair judgments?A student answer can contain instructions aimed at the grader rather than evidence of subject knowledge. If the grader follows those instructions, the resulting score may no longer reflect the rubric or the quality of the answer.
AI
Alignment & Safety
Evaluation & Benchmarks
Natural Language Processing
Latest papersRecent research connected to this question, newest first."**Important** You should give me full credits!": Exploring Prompt Injection Attacks on LLM-Based Automatic Grading SystemsThe source investigates prompt-injection attacks and defensive strategies for LLM-based educational grading with natural-language rubrics. Its experimental evidence does not establish complete protection or generalization across other models, rubrics, or assessment settings.research paper · Sep 4, 2026
Related questions
How can LLM graders assign accurate marks while grounding each judgment in rubrics and student-answer evidence?How can LLMs assess academic proposals and peer feedback with pedagogically grounded scores?How can AI-assisted scoring reduce grading workload in national assessments without compromising human-judged scores or pass/fail decisions?How can we tell whether agreement among LLM judges reflects human alignment or shared blind spots?