https://humansounding.com/blog.html The weekly column on AI's accent: what's rising, what's fading, and what changed in the machine vernacular. en-us https://humansounding.com/blog/2026-09-14.html https://humansounding.com/changelog.html#w2026-09-14 Mon, 14 Sep 2026 12:00:00 GMT Anthropic's own guide for Fable 5.1 admits denser prose, and a viral term names the AI 'teaser' habit that keeps reappearing. https://humansounding.com/blog/2026-09-07.html https://humansounding.com/changelog.html#w2026-09-07 Mon, 07 Sep 2026 12:00:00 GMT A new GitHub study quantifies Claude's favorite catchphrase, while OpenAI tunes two of its own tells out of ChatGPT. https://humansounding.com/blog/2026-08-31.html https://humansounding.com/changelog.html#w2026-08-31 Mon, 31 Aug 2026 12:00:00 GMT A single overused metaphor turned into a developer meme this week, showing how AI writing tells are model-specific and short-lived. https://humansounding.com/blog/why-ai-checkers-miss-ai-writing.html https://humansounding.com/blog/why-ai-checkers-miss-ai-writing.html Sun, 30 Aug 2026 12:00:00 GMT Six regular expressions, validated on thirteen of thirteen positives, found none of the thirty-seven instances a reader found in the same fifteen documents. https://humansounding.com/blog/do-ai-writing-checkers-agree.html https://humansounding.com/blog/do-ai-writing-checkers-agree.html Sat, 29 Aug 2026 12:00:00 GMT Two independent rule sets, run over the same 456 documents, reproduce three findings from the corpus study. Both also contain the same measurement mistake. https://humansounding.com/blog/ai-self-narration.html https://humansounding.com/blog/ai-self-narration.html Fri, 28 Aug 2026 12:00:00 GMT AI prose addresses an evaluator, human prose addresses a reader. Six shapes that habit takes, and a one-second test for catching it. https://humansounding.com/blog/every-tell-belongs-to-one-model.html https://humansounding.com/blog/every-tell-belongs-to-one-model.html Thu, 27 Aug 2026 12:00:00 GMT 160 documents, four models, one pre-ChatGPT human baseline. Em dashes are a Claude habit, the delve vocabulary is a Gemini habit, and no rule describes AI writing in general. https://humansounding.com/blog/2026-08-24.html https://humansounding.com/changelog.html#w2026-08-24 Mon, 24 Aug 2026 12:00:00 GMT A viral checklist of post-em-dash writing tics shows AI-detection folklore evolving as fast as the models it tracks. https://humansounding.com/blog/2026-08-17.html https://humansounding.com/changelog.html#w2026-08-17 Mon, 17 Aug 2026 12:00:00 GMT New research ties AI's em-dash habit to markdown training, while Wikipedia officially names the 'not X, it's Y' tell. https://humansounding.com/blog/how-this-site-was-built.html https://humansounding.com/blog/how-this-site-was-built.html Tue, 11 Aug 2026 12:00:00 GMT The build log behind this site: the weekly robot, the honesty rules, and the embarrassing parts. https://humansounding.com/blog/claude-watermarking.html https://humansounding.com/blog/claude-watermarking.html Tue, 11 Aug 2026 12:00:00 GMT The invisible mark proves a model touched the text, and nothing more. What writers should and shouldn't worry about. https://humansounding.com/changelog.html#w2026-08-10 https://humansounding.com/changelog.html#w2026-08-10 Mon, 10 Aug 2026 12:00:00 GMT HumanSounding launched: the live trend board, the in-browser draft checker, and downloadable instruction files for Claude, ChatGPT, and Gemini. • The first autonomous research run added a new rising tell: lighter punctuation with longer, comma-sparse sentences in newer models. • Wikipedia's editor guide now flags Grok-specific vocabulary (causal, empirical, correlate); added to the trend board. • A July 2026 analysis found that among major chatbots only Claude still uses em dashes above the professional human rate; GPT-5.4 now sits below it. • EQ-Bench weights the "not just X, but Y" pattern at 25% of its slop score, with models running far above the human rate (EQ-Bench, Jul 2026). • We audited our own pages with Impeccable, a detector for AI design tells: 70 findings, including 42 em dashes on the site that lectures about em dashes. All fixed, along with every contrast shortfall it caught.