SEBBI.PRO · Bias & Drift Test · Deep Check · Trust Centre

BIAS & DRIFT TEST

Is your AI fair? Prove it with numbers.

Drop in your decision log. Your phone or computer counts the outcomes for each group, so the records never leave your device. sebbi.pro works out the fairness measures regulators and courts use, tells you plainly what they mean, and with a key seals the result into Bitcoin-timestamped proof.

FOUR-FIFTHS RULE

Is any group selected at under 80% of the rate of the best-treated group?

SIGNIFICANCE

A chi-square test says whether the gap is real or could be chance.

EQUAL OPPORTUNITY

With true outcomes, compares how often each group is correctly approved or wrongly flagged.

DRIFT

Compare with an earlier period: has the population or any group's outcome moved?

1. Your decisions

A CSV file with one row per decision. It needs a column for the group (for example age band, sex or region) and a column for the outcome. A column with the true result is optional.

Or paste it:

2. The result

    Questions

    Does my data get uploaded?

    No. Your browser reads the file and sends only the counts for each group: how many decisions, how many favourable and, if you gave true results, how many were right or wrong. No names, no rows.

    Why arithmetic and not an AI?

    Fairness measures are formulas. Arithmetic gives the same answer every time, so anyone can recompute it from the sealed counts and get the same verdict.

    Which laws does this help with?

    EU AI Act Articles 10 and 15, the UK Equality Act 2010, UK GDPR fairness and Article 22, and the US EEOC four-fifths guideline. A FAIL is not a legal finding; it tells you where to look before a regulator does.

    Free to check. Sealing: 7 days free, then 50p per device per month.

    AI news
    TechCrunchThe AI graveyard: a running list of projects and startups that didn’t make it1hThe DecoderGemini 4 Carbon takes the MathArena crown but costs 15 times more than GPT-6.1 Sol2hThe DecoderAnthropic says outdated prompting habits are slowing down Claude Opus 5.52hTechCrunchAsian governments react to Nvidia CEO Jensen Huang’s exit2hThe VergeThe OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs3hWiredBeware the New AI-Fueled Ransomware Boom5hThe DecoderOpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI unnecessary7hThe DecoderAnthropic can already retrain Mythos with "a fraction of the compute originally needed," says former researcher Coxon7hThe DecoderRunway's open-source Praxis-1 turns video AI into a robot controller9hThe Guardian‘It’s like the wild west’: how AI has rocked the literary world9hTechCrunchJensen Huang’s departure blasts big hole in AI startup funding, VCs and founders say20hArs TechnicaFAA expands AI advice to manage air traffic congestion across entire US airspace21hThe VergeChatGPT Wallet lets OpenAI’s Dots agents make purchases for you22hBBC NewsPressure grows on Trump's Commerce Secretary Lutnick in fallout from Nvidia scandal22hThe GuardianNvidia scandal: US treasury chief blames Lutnick’s commerce department over alleged lack of action22hTechCrunchAI agent startup LangChain hits $3B valuation23hThe GuardianWhy the Nvidia scandal is far from over - The Latest23hTechCrunchExport-banned Nvidia chips are behind China’s most powerful AI models — uh-oh, Jensen!1dThe GuardianHow it feels to battle with a flood of AI-generated complaints | Letters1dArs TechnicaOpenAI board member worries an AI "takeover" of society could be months away1dThe Guardian‘Ahead of his time’: AI helps unlock discoveries hidden in notebooks of man who transformed maths1dThe VergeClaude’s Duet agent is a little neurotic and surprisingly helpful1dBBC NewsHow a sex scandal at Nvidia is putting pressure on the White House1dWiredThis Startup Is Betting on AI Microdramas to Get More Black Stories in Hollywood1dTechCrunchThe AI graveyard: a running list of projects and startups that didn’t make it1hThe DecoderGemini 4 Carbon takes the MathArena crown but costs 15 times more than GPT-6.1 Sol2hThe DecoderAnthropic says outdated prompting habits are slowing down Claude Opus 5.52hTechCrunchAsian governments react to Nvidia CEO Jensen Huang’s exit2hThe VergeThe OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs3hWiredBeware the New AI-Fueled Ransomware Boom5hThe DecoderOpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI unnecessary7hThe DecoderAnthropic can already retrain Mythos with "a fraction of the compute originally needed," says former researcher Coxon7hThe DecoderRunway's open-source Praxis-1 turns video AI into a robot controller9hThe Guardian‘It’s like the wild west’: how AI has rocked the literary world9hTechCrunchJensen Huang’s departure blasts big hole in AI startup funding, VCs and founders say20hArs TechnicaFAA expands AI advice to manage air traffic congestion across entire US airspace21hThe VergeChatGPT Wallet lets OpenAI’s Dots agents make purchases for you22hBBC NewsPressure grows on Trump's Commerce Secretary Lutnick in fallout from Nvidia scandal22hThe GuardianNvidia scandal: US treasury chief blames Lutnick’s commerce department over alleged lack of action22hTechCrunchAI agent startup LangChain hits $3B valuation23hThe GuardianWhy the Nvidia scandal is far from over - The Latest23hTechCrunchExport-banned Nvidia chips are behind China’s most powerful AI models — uh-oh, Jensen!1dThe GuardianHow it feels to battle with a flood of AI-generated complaints | Letters1dArs TechnicaOpenAI board member worries an AI "takeover" of society could be months away1dThe Guardian‘Ahead of his time’: AI helps unlock discoveries hidden in notebooks of man who transformed maths1dThe VergeClaude’s Duet agent is a little neurotic and surprisingly helpful1dBBC NewsHow a sex scandal at Nvidia is putting pressure on the White House1dWiredThis Startup Is Betting on AI Microdramas to Get More Black Stories in Hollywood1d