Guides · EU AI Act
The EU AI Act made the AI log a legal requirement.
For high-risk AI systems, the Act does not ask for a policy document. It requires the system to record what it did, automatically, over its whole lifetime. That record has to exist already. Vidai keeps it from the first call.
What it is
The first horizontal law for AI.
Regulation (EU) 2024/1689, the EU AI Act, is the first comprehensive law on how AI is built and used. It is risk-tiered: most AI carries light or no obligations, but a defined set of high-risk uses carries hard duties, phased in on a staged timeline through 2026 and 2027.
It reaches any organisation whose AI output is used in the EU, the same extraterritorial shape as the GDPR.
01
It is risk-scoped
The heavy obligations attach to high-risk systems, the Annex III categories, not to every AI call. Knowing which of your workflows are in scope is the first move.
02
Logging is mandatory
For high-risk systems, automatic event logging over the system's lifetime is not optional and not satisfied by a person reviewing outputs. It is a technical requirement.
03
You cannot backfill it
The record has to have been kept while the system ran. A log assembled after the fact is not the lifetime record the Act describes.
Source: Regulation (EU) 2024/1689, as published in the Official Journal. Dates for later stages can change; the official text decides.
Is your AI in scope
High-risk is a defined list.
- Creditworthiness and credit scoring of people
- Risk assessment and pricing in life and health insurance
- Recruitment, and decisions on promotion or termination
- Access to essential public and private services
- Critical infrastructure, education, law enforcement and migration
Most AI traffic is not high-risk
The Act says so. Because Vidai sits in the path, you don't have to decide per application whether to keep a record: it exists for all of it, and you scope the obligation to the workflows in Annex III.
Where Vidai fits
Vidai keeps the record the Act asks for.
The log is in the path, not after it
Every call is recorded as it happens: who, when, which model, what was refused and why. There is no separate logging step to remember.
Every AI system has a risk level
Each agent and application carries a risk level in the Act's terms, with an owner and a review date. A call by a high-risk system is always in the auditor's sample.
Oversight you can show
Approvals, overrides, suspended keys and Stop AI traffic are on the record, so human oversight is something you can point to, not just describe.
Precise about the boundary
Conformity assessment, the technical documentation and risk management stay your process. Vidai gives you the evidence; your auditor makes the call.
The mapping, article by article
Where Vidai's records serve the Act.
| Art. 12Record-keepingEvidence | High-risk systems must allow the automatic recording of events over their lifetime. Every call through Vidai is logged automatically: who, when, which model, what was refused and why. |
|---|---|
| Art. 19Automatically generated logsPart of it | Providers keep the logs under their control for at least six months, unless other law asks for longer. Request logs are the record; how long they're kept is your setting, and following the EU AI Act sets at least 180 days. |
| Art. 26(5)Monitoring high-risk systemsPart of it | Deployers monitor the operation of high-risk systems. Request logs and findings show how each agent behaves; informing the provider is yours. |
| Art. 26(6)Deployers keep the logsEvidence | Deployers keep the logs under their control for at least six months; financial institutions may do so through their existing record-keeping. Request logs and the audit log are kept by your installation. |
| Art. 14Human oversightPart of it | High-risk systems must be designed so people can oversee them. Stop AI traffic, model access, and suspending keys and agents are the controls; oversight itself is your process. |
| Art. 9Risk management systemPart of it | Risk levels record your assessment of each agent and application. The risk management system is your process. |
| Art. 5Prohibited AI practicesPart of it | A prohibited use is kept out with model access. Deciding what is prohibited is yours. |
Your organisation's process, not the product: AI literacy (Art. 4), data governance (Art. 10), the fundamental rights impact assessment (Art. 27) and transparency (Art. 50). Most high-risk duties fall on providers. The mapping is in beta: our reading of Regulation (EU) 2024/1689, not legal advice; the official text decides.
Proof · Risk register
Which AI systems are high-risk, and who owns them
Every agent and application with its risk level, owner, assessment date and review date. A call keeps the risk its system had when it was made.
On Enterprise, the Regulations page shows each article as enforced, observed, partly in place or a gap, with the calls that prove it, and a serious-incident register sits beside it.
How we start together
Walk through your high-risk AI logging.
- 01
Pick one app or team
Point it at the Control room. One configuration change; nothing else changes.
- 02
Watch for two weeks
Every call priced, checked and recorded. Rules start in observe mode.
- 03
See the evidence
What it cost, what it caught and what it would have stopped, from your own traffic.