UC San Diego AI Accusation After ADHD Writing Tool Use

Students with ADHD often rely on writing-support tools: text-to-speech readers, dictation software, outlining apps, and editors like Grammarly. When an AI detector flags a paper drafted with that kind of assistance, the accusation collides with disability accommodation in ways most academic integrity processes are not built to handle. The pattern is now appearing at UC San Diego, and it follows a shape worth understanding before you respond.

The UC San Diego pattern

UC San Diego routes academic integrity allegations through its Academic Integrity Office, which receives instructor reports and decides whether to pursue a charge. The familiar sequence: an instructor runs a paper through Turnitin's AI indicator (or a comparable tool), sees an elevated score, and reports a suspected violation. The student is contacted, often by email, and asked to respond.

What makes the ADHD-related cases distinct is the writing process. Students with ADHD frequently draft with the help of:

• Dictation tools such as Apple Dictation, Google Voice Typing, or Otter.ai, which produce cleaner, more uniform syntax than typed first drafts
• Text-to-speech readback tools that prompt the writer to smooth awkward phrasing during revision
• Grammar and clarity editors like Grammarly, which standardize sentence structure
• Outlining and focus tools that encourage short, linear paragraphs

Each of these tools tends to reduce the kind of variation that detectors read as human. None of them generates content. The student writes every idea and every sentence. The output simply looks statistically smoother than a draft typed straight into a blank document.

Why accommodation tools trigger AI detectors

AI detectors estimate two signals: perplexity (how predictable the word sequence is) and burstiness (how much sentence length and complexity vary). Writing produced with dictation and editing tools often scores low on both, because the writing process itself smooths out the irregularities that detectors associate with human authorship.

Independent researchers led by Debora Weber-Wulff tested fourteen AI detection tools in 2023 and concluded that none performed reliably enough across conditions to support institutional decisions. Stanford researchers (Liang et al., 2023, Patterns) separately found that detectors disproportionately flag writing that uses simpler syntax and common vocabulary, which is exactly the pattern many assistive workflows produce. The same issue surfaces in Grammarly false positive cases and in writing from non-native English speakers.

What UC San Diego policy actually requires

UC San Diego's Policy on Integrity of Scholarship prohibits unauthorized assistance and misrepresentation of authorship. It does not, as a blanket matter, prohibit assistive technology. Authorized accommodations are administered through the Office for Students with Disabilities (OSD), and instructors are expected to follow accommodation plans on file.

Two procedural points matter in these cases:

1. An instructor's suspicion is the starting point, not the finding. The Academic Integrity Office reviews the report and the student's response before the matter proceeds.
2. A detector score is not, by policy, sufficient evidence of a violation on its own. The instructor must connect the score to a specific claim about what the student did.

Evidence that shifts these cases

When ADHD-related assistive workflows are part of the picture, the evidence that tends to close the file falls into three categories:

• Process evidence: Google Docs or Microsoft Word version history showing the document built up over time, dictation transcripts with timestamps, and notes or outlines that predate the draft.
• Accommodation documentation: Your OSD accommodation letter, any correspondence with the instructor about approved tools, and a description of how the tools are used in your writing workflow.
• Subject-matter familiarity: Annotated sources, your notes on the readings, and the ability to explain the argument in your own voice if asked.

Detector inconsistency also helps. Running the same paper through multiple tools (Turnitin, GPTZero, Originality.ai, Copyleaks) and documenting disagreement among them undercuts the reliability of any single score. The research on detector accuracy is directly citable here.

If this is you at UC San Diego

Read the email from your instructor or the Academic Integrity Office carefully. Note the deadlines and the exact language of the allegation. Request, in writing, the specific detector used, the score reported, and any human review notes. Preserve every draft, transcript, outline, and source file from your writing process. If you have an OSD accommodation, attach the documentation and explain in plain language how each tool you used works and why you use it.

Your written response is the single most important document in the process. It needs to do three things at once: address the detector score on its own terms, document your writing process, and connect your tool use to your accommodation. NotBot generates a personalized defense package that does this work in about a minute, including a response letter, an evidence guide, and a hearing prep brief tailored to the detector and the allegation. If you are past a finding and considering an appeal, the procedural grounds shift, and the procedural rights FAQ covers what to ask for at that stage.

If the proposed sanction includes suspension, dismissal, or anything that would affect financial aid or visa status, consult an education law attorney before your meeting. Accommodation-adjacent cases sometimes raise disability discrimination questions that benefit from legal review.