Short-form writing
Short posts give AI detectors less evidence and readers more reason to be cautious.
A short post may contain only a few sentences, a slogan, or a common factual construction. That leaves less linguistic context for a detector and makes a confident origin claim especially difficult to justify.
Small sample
A caption or post may not contain enough varied prose for a stable classification about its origin.
Conventional language
Headlines, announcements, calls to action, and factual updates often reuse predictable structures across human and generated text.
Process fits the use case
Recording the post as it develops provides evidence tied to that text without pretending to classify its style.
Why length matters
A classifier needs enough signal to distinguish patterns from ordinary variation. With only a few words or sentences, punctuation, topic, quoted material, and platform conventions can dominate the sample. Expanding the text with unrelated material would change the question rather than improve the evidence.
Some institutional detectors explicitly require long-form prose and minimum word counts. That limitation should be visible whenever someone proposes scanning a tweet, headline, product description, email subject, or short caption.
Editing makes the origin question harder
Short text is often revised heavily. A human may compress a paragraph into one polished sentence; generated copy may be rewritten until little of the original phrasing remains. The final output alone cannot reveal that mixed history with certainty.
Translation, house style, templates, and legal wording add further ambiguity. A predictable sentence may reflect a constraint rather than a machine, while unusual language may simply reflect an individual writer.
What to use for a public post
If process transparency matters, record the writing while the post is created. Keep relevant notes and sources, disclose assistance that affected the wording, and attach a receipt that explains exactly what its controlled session did and did not observe.
Do not convert the receipt into a “human” badge. HumanTyped can reconstruct the sealed post from accepted edits and present the session evidence; it still cannot observe another device, identify the writer, or determine the origin of every idea.
Primary sources
Check the current product and evidence record.
Evidence, not certainty