Dealing with AI-Generated Cheating in High School

The release of ChatGPT and similar large language models has fundamentally shifted the academic environment. High school administrators and teachers are no longer just looking for copy-pasted text from Wikipedia; they are now facing unique, generated responses that bypass traditional plagiarism checkers. This shift has forced districts to rapidly update honor codes, invest in new detection technology, and rethink how they assess student learning.

The Shift in Academic Honor Codes

For decades, academic integrity policies focused on “unauthorized assistance” and “plagiarism.” However, the definition of plagiarism has become murky. Technically, AI-generated text is unique and not stolen from a specific author, which creates a loophole in older rulebooks.

To close this gap, school districts across the country are rewriting their handbooks to specifically address “Generative AI.”

From Bans to Integration

Initially, the knee-jerk reaction was a total ban. For example, New York City Public Schools, the largest district in the United States, blocked ChatGPT on school devices and networks in January 2023. However, by May 2023, the district reversed this decision. They realized that prohibition was impossible to enforce once students went home to their personal devices.

Now, updated honor codes typically fall into three categories:

  • Total Prohibition: Any use of AI to generate ideas, outlines, or text is considered a violation of academic integrity.
  • Disclosure Required: Students may use AI tools for brainstorming or editing but must cite the AI model used and provide the prompt history.
  • Teacher Discretion: The policy allows individual teachers to set the rules for specific assignments. A creative writing teacher might ban it, while a computer science teacher might encourage it for coding assistance.

The Reality of Detection Software

As students adopted AI tools, educational technology companies rushed to provide solutions. Schools are currently spending significant portions of their budgets on software that claims to identify machine-written text.

Major Players in AI Detection

  • Turnitin: This is the industry standard for plagiarism checking. In April 2023, they activated an AI detection feature available to thousands of high schools and universities. Turnitin analyzes the predictability and pattern of words to assign a probability score.
  • GPTZero: Created by Edward Tian, a Princeton student, this tool specifically looks for “perplexity” (randomness) and “burstiness” (variation in sentence structure) to flag AI content. Many teachers use the free version or paid district licenses.
  • CopyLeaks: This enterprise-level solution claims high accuracy and is used by various educational institutions to scan for content generated by GPT-4 and other models.

The Problem of False Positives

Despite marketing claims of 98% or 99% confidence, these tools are not foolproof. This unreliability creates a massive headache for administration.

A 2023 study by researchers at Stanford University highlighted a worrying trend: AI detectors are biased against non-native English speakers. The study showed that detectors flagged over half of the essays written by non-native English speakers as “AI-generated” because their writing tends to be simpler and more predictable.

Consequently, many universities (such as Vanderbilt and Michigan State) turned off Turnitin’s AI detector. High schools are following suit, often using the software as a “flag” for further investigation rather than definitive proof of guilt.

Changes in Classroom Assessment Strategies

Because detection software is unreliable, teachers are reverting to analog methods to ensure students are actually doing the work.

The Return of the “Blue Book”

Many AP English and History classes are moving back to in-class, handwritten essays. By requiring students to write in a “Blue Book” under supervision, teachers can verify that the work is authentic. This eliminates the possibility of using a chatbot during the drafting process.

Version History Forensics

Instead of relying on AI detectors, savvy teachers are using the “Version History” feature in Google Docs. This allows them to see a replay of the student’s typing.

  • Natural writing: Shows pauses, backspaces, edits, and gradual text accumulation over hours or days.
  • Cheating: Often appears as large blocks of text pasting in instantly, or a document that is created and finished in five minutes.

Oral Defenses

Some districts are encouraging “oral defense” of papers. If a student submits a complex essay, the teacher may pull them aside and ask them to explain specific vocabulary choices or arguments used in the paper. If the student cannot explain what they “wrote,” it serves as evidence that they did not generate the thoughts themselves.

Protecting Students from False Accusations

With the rise of aggressive detection, students need to protect themselves from false accusations of cheating. Parents and guardians should advise high schoolers to take the following precautions:

  1. Work in the Cloud: Always write essays in Google Docs or Microsoft Word Online. Do not write in a separate note app and paste it over. You need the timestamped version history to prove you typed it manually.
  2. Save Drafts: Keep rough notes, outlines, and early drafts. If a teacher questions the final product, showing the evolution of the idea is the best defense.
  3. Screen Record: For high-stakes assignments, some students now use screen recording software to document the actual writing process.

Frequently Asked Questions

Can Turnitin prove 100% that I used ChatGPT?

No. Turnitin provides a probability score, not definitive proof. They openly state that their tool should not be the sole basis for disciplinary action. However, a high score can lead to a meeting with administration where other evidence (like lack of version history) is reviewed.

Is using Grammarly considered AI cheating?

It depends on the specific school policy and the version of Grammarly. The standard spell-check is usually fine. However, Grammarly GO and other “generative” features that rewrite whole sentences or paragraphs trigger AI detectors and often violate strict honor codes.

What happens if I am caught using AI?

Penalties vary by district. Common consequences include receiving a zero on the assignment, mandatory redo of the work for reduced credit, suspension, or a permanent mark on the student’s disciplinary record. For National Honor Society members, an academic integrity violation usually results in immediate dismissal from the organization.

Do AI detectors work on translated text?

If a student writes in their native language and uses Google Translate to convert it to English, it often gets flagged as AI. This is because translation tools use similar predictive algorithms to LLMs (Large Language Models). Students should clarify translation policies with their ESL teachers.