About This Agency
A brief institutional overview for those who require one. For those who do not — the detector is this way.
“AIGENERATEDBULLSHIT.COM is a fully operational, AI-powered bullshit detection agency committed to transparency, accountability, and the ruthless measurement of corporate non-communication. We take our work extremely seriously. The irony is not lost on us.”
— The AIGBS Director General, Est. 2026
The BSI — Bullshit Index
Every result produced by this tool is expressed as a BSI — Bullshit Index score. A single number, 0 to 100, representing the concentration of corporate non-communication in any given text.
The BSI is calculated by scoring the submitted text across 10 proprietary detection categories — each measuring a distinct dimension of corporate non-communication — and computing a weighted overall index. A BSI of 0 means the text says something real. A BSI of 100 means it says nothing at all, with great confidence.
The BSI is designed to be shareable. When you share a result, the format is:
The grade mirrors academic notation — A through F — with F reserved for text that has achieved a kind of negative excellence. A perfect BSI of 100 has not yet been recorded. We remain hopeful.
Our Mission
Every day, approximately 300 billion corporate emails are sent worldwide. Studies suggest that a significant proportion of them say nothing whatsoever. The studies themselves may also say nothing — we have not checked. What we do know is that the volume of professionally produced text that communicates nothing has never been higher, and nobody was measuring it.
AIGENERATEDBULLSHIT.COM was founded on a simple premise: someone should measure this. Not to fix it — fixing it is above our pay grade and frankly above everyone else's too — but to quantify it. To look a press release in the eye and say: BSI = 87%, Grade F, and we can prove it. To give corporate non-communication the number it deserves and the public shaming it has earned.
We believe in radical transparency. Specifically, we are radically transparent about how little transparency exists in modern corporate communication. Every submission is scored consistently, every category is explained, and every result is shareable. If your email scores 94% on the Bullshit Index, that information belongs in the world.
How It Works
Our detection engine is powered by Claude, Anthropic's AI model — which means we are using AI to detect AI-sounding language, a situation we find professionally satisfying and slightly poetic. Text is submitted through the detector, pre-classified by genre, and then scored across 10 proprietary bullshit dimensions. The entire process takes approximately five seconds and costs the user nothing.
Each of the 10 categories measures a distinct dimension of corporate non-communication — from Synergy Fluff and Fake Urgency to the Confidence–Evidence Gap and the Euphemism Body Count. Scores run from 0 to 100 per category, producing an overall BSI expressed as a percentage and a letter grade from A to F. Grade F is not uncommon. Grade A is rarer than it should be.
Synergy Fluff
Buzzwords that commit to nothing
Fake Urgency
Pressure with no real deadline
Responsibility Fog
Nobody is accountable for anything
Metric Abuse
Numbers that mean nothing
Startup Dialect
Silicon Valley vocabulary density
Virtue Signalling
Posturing with no action attached
Padding & Filler
Words that add length, not meaning
Confidence–Evidence Gap
Bold claims, zero proof
AI Fingerprint
AI pretending to be human
Euphemism Body Count
Bad news dressed as opportunity
Scoring is calibrated using 10 anchor texts — reference submissions with known correct scores that appear in every analysis request, giving the AI a consistent internal ruler. Temperature is set to zero, ensuring that the same text submitted twice always returns the same score. Repeat submissions are served from cache instantly. The science is real. The satire is intentional. Both are non-negotiable.
Classic Mode
Not everything that is unclear is corporate. Some of it is just Kafka. Our AI pre-screens submitted text for genuine literary or creative writing — poetry, fiction, prose of genuine merit, dramatic works, and canonical literature. If detected, standard BS scoring is suspended and Classic Mode is activated, because measuring Shakespeare against corporate communication standards felt unfair to Shakespeare.
In Classic Mode, the text receives a satirical academic assessment using genre-appropriate alternative metrics — Melodrama Density, Orphan Deployment Rate, Iambic Compliance, Narrator Unreliability Score, and others — issued by the AIGBS Institute of Classical Studies, Est. 2026. The scores are entirely fabricated. The verdict is always Certified Non-Bullshit. The apology to literature is sincere.
Classic Mode was introduced after someone submitted Hamlet and our detector gave it a 94% Bullshit Score. We felt we owed literature an apology. We also felt we owed it a funnier one. It also felt unfair to corporate communication, which has enough problems without being compared unfavourably to Elizabethan theatre. Classic Mode resolves both injustices simultaneously.
FAQ
Q: Is this accurate?
A: The scores are consistent and calibrated but should not be used to make employment decisions, file lawsuits, or win arguments with your manager. They reflect linguistic patterns associated with corporate non-communication — not a moral judgement. They are, ironically, somewhat bullshit themselves, in the sense that a number always implies more precision than actually exists. We are at peace with this.
Q: Can I submit my own writing?
A: Yes. The detector will score it honestly. We accept no responsibility for what it finds, what it implies about your communication style, or any resulting crisis of professional identity. Many users report that seeing their own writing scored is clarifying. Several have reported it is deeply uncomfortable. Both responses seem appropriate.
Q: What if my text scores very high?
A: Congratulations. You have achieved something genuinely rare — text so thoroughly devoid of meaning that our AI was momentarily impressed. A score above 90% puts you in the Weapons-Grade Bullshit tier, occupied by the finest press releases, political non-answers, and startup pitch decks humanity has produced. Consider a career in investor relations, political communications, or corporate strategy. The market is robust.
Q: What if my text scores very low?
A: You communicate clearly and directly. This is statistically unusual. We are slightly suspicious. A score below 20% means our detector found almost nothing to flag — no padding, no fake urgency, no responsibility fog. Either you write with genuine clarity and specificity, or you have discovered a way to encode bullshit that our system cannot yet detect. We are working on both hypotheses.
Q: Can I submit emails from my boss?
A: We cannot stop you. We also cannot be held responsible for what you do with the results, whether you share them, whether you print them and leave them on a desk, or whether they inform any professional decisions. We are a satirical tool. What happens between a user and their BSI score is between them and their conscience.
Q: Is the AI Generated Bullshit Detector itself bullshit?
A: Our lawyers have advised us not to answer this question. Our AI, when asked to score our own About page, returned a BSI of 31% — Mildly Suspicious. We consider this a reasonable result and a personal best.
Q: Why does Classic Mode exist?
A: Someone submitted Hamlet. The AI gave it a 94% Bullshit Score on the grounds of excessive melodrama, responsibility fog (nobody accepts accountability for anything in Hamlet), and a confidence-evidence gap in the ghost's claims. We felt we owed literature an apology. Classic Mode is that apology.
Q: How often is the scoring updated?
A: The scoring prompt is anchored to 10 calibration texts and pinned to a specific model version, so scores are stable over time. When we update the model or recalibrate anchors, previously cached results are unaffected — they return the same score they always did. We treat consistency as a core feature, not an accident.
Who Built This
AIGENERATEDBULLSHIT.COM is an independent project — not affiliated with any corporation, consultancy, political party, or organisation whose communications might otherwise appear in our database. It was built by a single developer as a pilot project with a budget of approximately what a decent dinner costs and a timeline of three months. This is relevant context for the disclaimers that follow.
The technical infrastructure runs on Vercel, Supabase, Upstash, Resend, and Anthropic's Claude API. The domain name was available. We took that as a sign. The site is ad-supported and costs approximately nothing to run, which is appropriate for a project about things that cost a great deal of effort while producing very little of value.
The BS Detector bookmarklet extends the tool to any webpage — install it once and score any text on the internet in two clicks. The weekly BS Trend Report delivers the week's finest corporate non-communication to your inbox every Monday, generated entirely by AI, requiring no human editorial input whatsoever. This is either a feature or a warning, depending on your perspective on AI-generated content about AI-generated content.
Disclaimer
AIGENERATEDBULLSHIT.COM is a satirical tool for entertainment and educational purposes. BS scores are generated by AI and reflect the model's assessment of linguistic patterns statistically associated with corporate non-communication. They do not constitute professional advice of any kind — legal, financial, psychological, or otherwise. We are not responsible for any employment decisions, relationship breakdowns, or existential crises that result from a high BSI score.
Any text submitted may have been written by a perfectly reasonable person doing their best in a difficult institutional context, constrained by style guides, legal review, or a manager who insisted on the word “leverage”. We are not judging the author. We are judging the text. There is a difference. Probably.
The scores, categories, and grades produced by this tool are designed to be consistent, calibrated, and entertaining — not definitive. If your organisation uses this tool to make decisions about employees, communications, or strategy, that is your business. We would gently suggest that a satirical AI bullshit detector should not be the primary input to serious institutional decisions. We would also note that a press release scoring 91% probably deserves a rewrite regardless of what any tool says.
You have read the About page. You are now legally required to submit something.
Detect Some Bullshit