The AI Act asks high-risk systems to keep logs and asks many AI systems to tell people they are talking to a machine. Trustvice does not make you compliant. It gives the records you keep a signature and a time that you cannot change later.
TrustActorPlacardNotarianTOPSLensLogging is an engineering task; showing that the logs are complete and unaltered years later is a different one. A record that only you can vouch for is weaker evidence than one a third party can verify.
Can we show what the system did on this request, in order?
How do we prove the AI notice was actually delivered in German, not just written?
Can we show this record was not edited after the question was raised?
TrustActor keeps agent runs in a tamper-evident hash chain, Placard signs the AI-disclosure notices you serve and keeps a ledger of their delivery, and Notarian signs snapshots of the audit chain. You keep the content and the logs themselves; these records make them tamper-evident and dated.
AppliesIn force since 1 August 2024; general application from 2 August 2026
The AI Act sets obligations by risk level, with staggered dates. Its record-keeping and transparency rules are the ones closest to what Trustvice records.
Official text ↗AppliesAnnex III systems from 2 December 2027; Annex I systems from 2 August 2028
High-risk AI systems must technically allow the automatic recording of events (logs) over their lifetime, so that their operation can be traced: situations that may present a risk or a substantial modification, post-market monitoring, and monitoring of operation by deployers.
Official text ↗AppliesWith the high-risk obligations
Providers and deployers must keep the automatically generated logs under their control for a period appropriate to the system’s intended purpose, of at least six months, unless other Union or national law provides otherwise.
Official text ↗AppliesFrom 2 August 2026
Providers must design AI systems that interact directly with people so that those people are informed they are interacting with an AI system, unless it is obvious. For generative systems placed on the market before 2 August 2026, the Digital Omnibus on AI sets a transition period for the marking duty in Article 50(2); check its end date in the published text of Regulation (EU) 2026/1744.
Official text ↗AppliesIn force since 27 July 2026
Amended the AI Act’s timetable: the high-risk obligations now apply from 2 December 2027 for Annex III systems and from 2 August 2028 for systems covered by Annex I, instead of 2 August 2026.
Official text ↗Legal references are summarised and dated as of October 2026. They are given so you can see where the need comes from, not as legal advice. Read the official text, and ask counsel how it applies to you. Trustvice does not make any organisation compliant with any law.
Your system posts agent runs to TrustActor and, when an AI interaction starts, records which disclosure notice it delivered, on which channel and in which language.
Runs land in the workspace hash chain. Placard composes a signed, content-addressed AI-disclosure notice and records each delivery in its ledger.
Notarian snapshots of the audit chain and Lens evidence bundles give you signed exports for a period. Keep your own copy for as long as your record-keeping requires.
Anyone you share the records with can check snapshot anchors on the public verify path, and resolve and verify a Placard notice on its public endpoints, without an account.
Each suite covers one part of the record. Open a suite to see what it does.
Records each agent run in the workspace hash chain, with replay and diff, as the event log behind your own logs.
Open suite →Signed AI-disclosure labels with public resolve and a delivery ledger, for the transparency duty in Article 50.
Open suite →A capability of Trustvice 360: Ed25519-signed snapshots of the audit chain, so the records as they stood on a date can be checked later.
Open suite →A capability of Trustvice 360: the append-only transparency chain, sealed hourly and anchored with OpenTimestamps.
Open suite →Queries the audit chain for a period and produces a signed evidence bundle for an assessor.
Open suite →Two records an AI system can produce for a chat assistant: the AI-disclosure notice itself, composed once and signed, and a delivery record each time a session starts.
The question{
"workspaceId": "ws_…",
"subject": "support-assistant",
"channels": ["chat"],
"texts": [{
"locale": "de",
"elements": {
"artificialNature": "Sie sprechen mit einem KI-Assistenten, nicht mit einem Menschen."
}
}]
}
→ { "ok": true, "data": { "placardId": "plc_…", "version": 1, "signature": "…", … } }{
"workspaceId": "ws_…",
"placardId": "plc_…",
"channel": "chat",
"subjectRef": "sess_123",
"priorTurns": 0,
"locale": "de"
}
→ { "ok": true, "data": { "deliveryId": "plcd_…", "localeServed": "de", "localeFallback": false, … } }Request fields as documented: /docs/api/placard
A signed record is evidence of a narrow fact: these values were recorded by Trustvice at this time, and have not changed since. It is not a certificate, an audit, a qualified timestamp or legal advice, and it does not by itself satisfy any legal obligation.
Every workspace can reach every suite; plans differ by capacity. Read the docs, or talk to us about what you need to prove.
Agents, evidence, governance, commerce and disclosure are all in Trustvice 360. Sign up free and open every suite from one dashboard.