Definition
AI Sentence DNA is a term coined by Bloomberry Research to describe the compound pattern of co-occurring signals that makes AI-generated writing structurally recognizable. No single word or sentence shape is diagnostic on its own — AI Sentence DNA refers to the stacking of vocabulary elevation, cadence uniformity, formulaic hook constructions, smooth filler transitions, and predictable paragraph symmetry across a piece of text.
The term was introduced in Bloomberry's AI Sentence DNA corpus research — a database of 7,400+ catalogued AI-writing signal entries assembled from production enforcement systems, external academic research, and cross-model output analysis. The corpus is designed to catalogue recurring AI-writing patterns for use in voice calibration and writing quality systems, not to make authorship determinations.
AI Sentence DNA is the empirical foundation for Bloomberry's real-time writing screening systems. When Bloomberry generates a post, it runs the output through the live signal corpus — flagging vocabulary markers, scanning for cadence patterns, and detecting hook formulas — before calibrating the result against the user's individual voice profile.
AI Sentence DNA is not a single marker — it is a compound of five measurable signal layers that co-occur in AI-generated writing.
A measurable increase in the frequency of certain abstract, formal, or hedge words: delve, showcase, underscore, pivotal, leverage, navigate, tapestry, foster. These words are not AI-exclusive, but their co-occurrence at elevated density is a reliable compound signal. Kobak et al. (2025) documented post-ChatGPT vocabulary frequency increases reaching 28× baseline in academic abstracts.
AI models default to consistent sentence rhythm — balanced clause lengths, parallel constructions, and predictable 3- to 4-beat sentence progressions. The AI Sentence DNA corpus catalogs 12 named structural cadence detectors, including the Rhetorical Contrast cadence ("It's not just about X. It's about Y."), the Generic Opener, and the Resolution Closer.
AI writing tends toward predictable first-line constructions: generalization openers ("Most people..."), observer openers ("I've been thinking about..."), interrogative hooks ("What if you could...?"), and binary contrast openers. The corpus catalogs 17 named hook patterns. When multiple hook patterns appear across multiple paragraphs, the compound signal strengthens.
AI-generated prose uses high-information-density connective phrases that signal organization without adding meaning: "It's important to note," "Furthermore," "In today's fast-paced landscape," "As we move forward." These transitions act as structural glue but carry no substantive content. Their presence at elevated density is a consistent soft signal.
At the paragraph level, AI writing tends toward uniform block sizes, consistent internal structure (claim → expansion → example → conclusion), and balanced visual layout. Human writing produces more asymmetric structures: abrupt endings, unresolved tensions, paragraphs that concede more than they recover.
The term AI Sentence DNA was developed internally at Bloomberry to describe the layered signal pattern that Bloomberry's output screening systems were built to detect and avoid. The DNA analogy was chosen deliberately: just as DNA encodes structural information at a molecular level, these writing patterns encode recognizable structural information at a sentence level — patterns that persist across topic changes, model versions, and surface-level vocabulary variation.
The concept was formalized in Bloomberry Research Vol. 1 (The Emergence of AI Dialects) and expanded into a full corpus methodology in the AI Writing Patterns Database. As of June 2026, the corpus contains 7,622 audited signal entries across six entry types.
Related research: why AI sentence structure is structurally predictable · the 12 named AI sentence cadence patterns.
Rewrite in your voice with Bloomberry
Bloomberry detects AI Sentence DNA in real time and rewrites content against your individual voice profile — not a generic alternative.
Related research
AI Writing Patterns Database
The full corpus: 7,400+ catalogued AI-writing signal entries across vocabulary, cadence, structure, and style.
The Emergence of AI Dialects
Why AI models developed distinct writing dialects — and the AI Sentence DNA framework origin.
AI Sentence Structure
Why AI-generated sentences feel structurally predictable — the architectural blueprint.
AI Sentence Patterns
12 named cadence patterns from the corpus, with examples and rewrites.
How to Spot AI Writing Patterns
A practical guide to identifying co-occurring signals in AI-generated text.