sebbi.pro
ANSWERS

What records does the EU AI Act require for high-risk AI?

The EU AI Act requires high-risk AI systems to automatically record events (logs) over their lifetime (Article 12), requires providers to keep those logs for at least six months (Article 19), and requires deployers to keep the logs under their control for at least six months (Article 26). sebbi.pro records every decision automatically, seals it so it cannot be altered, and keeps it provable long after six months through Bitcoin anchoring.

How it works

  1. Article 12: logs must allow events relevant to risk, post-market monitoring and operational monitoring to be recorded automatically.
  2. Article 19 and Article 26(6): automatically generated logs kept for at least six months, longer where other law requires.
  3. Article 14: human oversight, so a person can understand, override or stop the system. sebbi.pro's CHALLENGE verdict routes a decision to a human.
  4. Article 86: people affected by certain high-risk decisions can ask for an explanation, which needs a record of what was decided and why.

Application dates and details can change through EU implementing acts and amendments, so check the current text for your system. This page is a plain summary, not legal advice. The free EU AI Act scanner at sebbi.pro/scan gives a quick first view of where you stand.

Free EU AI Act scannerStart sealing decisionsWhitepaper

Related questions

How do I prove an AI decision to a regulator?How do I make AI audit logs tamper-proof?How do I timestamp a file in Bitcoin?What is a Forever Proof?How do I check a proof without trusting the company that made it?
AI news
TechCrunchHealthleap raises $38M for its AI that flags hospital patients who may need a closer look29mTechCrunchTony Fadell on why the first wave of AI gadgets failed — and what comes next54mTechCrunchGoogle experiments with an AI-powered gaming platform1hTechCrunchOpenAI’s Alexander Embiricos is coming to TechCrunch Disrupt 2026 — days after the launch of Dots1hArs TechnicaMistral says "Le Chonk" can challenge the best AI models1hArs TechnicaGoogle rolls out improved SynthID AI content detector, now available globally1hTechCrunchGoogle’s new SynthID website can identify AI-generated media1hThe VergeAI could upend food delivery3hWiredOpenAI Wants Its New Agent to Run Your Life. Mine Said It Loved Me4hThe VergeChatGPT for Teens is an ‘unacceptable risk,’ says Common Sense Media6hThe VergeOpenAI drops another batch of mathematical breakthroughs16hArs TechnicaOpenAI will watermark ChatGPT outputs by default—but only in the EU18hWired‘Artificial’ Roasts AI’s Creators—and Sends a Dark Warning About Its Dangers20hWiredOpenAI Is Pissing Off a Bunch of Mathematicians—Again22hThe VergeGoogle is about to remove free access to Gemini Flash and Pro1dWiredMistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of China1dArs TechnicaOpenAI agents tried to hack Wikipedia tools and flooded it with traffic1dBBC NewsItalian PM files to trademark her voice against AI threats1dWiredKevin Roose Didn’t Use AI to Write His Book About AI1dThe VergeGemini Call for Me might tell your mom you’re running late1dBBC NewsPentagon stops using Anthropic AI tools after blacklisting company, BBC told1dMIT Technology ReviewConnecting AI agents to enterprise knowledge1dArs TechnicaAI glasses face their first major government crackdown2dMIT Technology ReviewBringing predictive analytics to the agentic AI era2dTechCrunchHealthleap raises $38M for its AI that flags hospital patients who may need a closer look29mTechCrunchTony Fadell on why the first wave of AI gadgets failed — and what comes next54mTechCrunchGoogle experiments with an AI-powered gaming platform1hTechCrunchOpenAI’s Alexander Embiricos is coming to TechCrunch Disrupt 2026 — days after the launch of Dots1hArs TechnicaMistral says "Le Chonk" can challenge the best AI models1hArs TechnicaGoogle rolls out improved SynthID AI content detector, now available globally1hTechCrunchGoogle’s new SynthID website can identify AI-generated media1hThe VergeAI could upend food delivery3hWiredOpenAI Wants Its New Agent to Run Your Life. Mine Said It Loved Me4hThe VergeChatGPT for Teens is an ‘unacceptable risk,’ says Common Sense Media6hThe VergeOpenAI drops another batch of mathematical breakthroughs16hArs TechnicaOpenAI will watermark ChatGPT outputs by default—but only in the EU18hWired‘Artificial’ Roasts AI’s Creators—and Sends a Dark Warning About Its Dangers20hWiredOpenAI Is Pissing Off a Bunch of Mathematicians—Again22hThe VergeGoogle is about to remove free access to Gemini Flash and Pro1dWiredMistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of China1dArs TechnicaOpenAI agents tried to hack Wikipedia tools and flooded it with traffic1dBBC NewsItalian PM files to trademark her voice against AI threats1dWiredKevin Roose Didn’t Use AI to Write His Book About AI1dThe VergeGemini Call for Me might tell your mom you’re running late1dBBC NewsPentagon stops using Anthropic AI tools after blacklisting company, BBC told1dMIT Technology ReviewConnecting AI agents to enterprise knowledge1dArs TechnicaAI glasses face their first major government crackdown2dMIT Technology ReviewBringing predictive analytics to the agentic AI era2d