I see this allot, concerns over being accused of AI. When I first published my childhood writings, for fun I ran it through an AI “Detector” It said 91% was AI. Current research describes detectors as vulnerable to false positives and unreliable when used punitively. I will not discuss the benefits or pitfalls of AI. Like computer use and the controversy over screens, everyone has a point the want to make. However, if you are concerned of being accused, here is some simple advice:

To protect yourself, focus on provenance, not detector scores. Keep drafts, notebooks, scans, dated files, handwritten pages, old emails, workshop notes, and early exports. Use version history: Word/Google Docs/OneDrive, Scrivener snapshots, Git, or a writing app that keeps dated revisions. Save process evidence: outlines, fragments, alternate lines, cut stanzas, handwritten edits. For older work, scan originals and keep photos of notebooks or folders with visible context. Keep a short AI-use statement: for example, “No generative AI was used to draft this work. Spellcheck/grammar tools, if any, were limited to proofreading.”
If submitting somewhere, check their AI policy before submission and keep a copy. If accused, ask for the detector name, version, date, threshold, full report, and whether their policy allows detector-only punishment. Then provide your draft trail and request human review.

Do not ruin your style trying to “sound less AI.” That punishes your own voice. The better defense is a paper trail showing how the work came into being. I have a website of my drafts and poetry, and I scanned in my old notebooks and drafts, in part to show process and also prove that I am the author of what I write. Yet I have also experimented with many forms, such as search bar poetry, auto-fill, and other technological experiments. My conscience is clean.