# Transparency Evidence Pulse

Canonical page: <https://hogarmas.net/transparency-pulse/>

Transparency Evidence Pulse is a free, browser-local change monitor for observable AI transparency evidence on a public website homepage. It creates a minimized baseline from the existing H/M AI Transparency Surface Audit, rescans the same origin later, and reports normalized evidence changes.

## Workflow

1. Enter one public HTTP or HTTPS website.
2. H/M performs a bounded static homepage scan. Local, private, reserved, credentialed, and non-standard-port targets are blocked.
3. The browser normalizes the scan into a baseline and creates a SHA-256 checksum.
4. Download the baseline. H/M does not store it for you.
5. Load that baseline later and rescan the exact same origin.
6. The browser produces a factual evidence delta and a second SHA-256 checksum.

The page also includes a clearly labeled synthetic example. Synthetic activity is never presented as a real website scan.

## Compared evidence

The current ruleset compares eight bounded evidence areas:

- integration signals in public markup
- AI interaction wording candidates
- synthetic-content wording candidates
- machine-readable marker candidates
- sensitive-use wording candidates
- accessible disclosure surface candidates
- first-interaction timing review status
- media-level provenance review status

It also compares fixed integration tokens, fixed machine-marker types, and four bounded candidate counters. It never returns or exports fetched source bodies.

## Decisions

- `no-observed-change`: no normalized difference was observed under the same evidence method.
- `change-observed`: at least one normalized evidence status, count, basis, fixed token, or counter changed.
- `method-changed`: the evidence version changed, so direct comparability requires human review.

These states do not mean compliant, non-compliant, improved, regressed, safe, authentic, or certified.

## Accepted local files

- `hm.transparency-evidence-baseline.v1`
- `hm.ai-transparency-surface-audit.v1`, which the browser can wrap into a new baseline
- JSON only, up to 512 KB

Checksums are verified before a baseline can be compared. A current scan must have the exact same canonical origin as the baseline.

## Export contract

Schema: <https://hogarmas.net/transparency-pulse/schema.json>

Exports use:

- `hm.transparency-evidence-baseline.v1`
- `hm.transparency-evidence-diff.v1`
- ruleset `2026-08-03.1`

The SHA-256 checksum detects changed canonical export bytes. It is not a digital signature and does not authenticate an issuer, person, deployment, or company.

## Privacy and metrics

- Fetched source content stored: false
- Fetched source content returned: false
- Baseline stored by H/M: false
- Diff stored by H/M: false
- Browser storage used: false
- Media downloaded: false
- Dynamic widgets executed: false
- Model called: false

Aggregate product metrics use fixed action identifiers only. Metrics never contain the scanned URL, evidence values, tokens, counters, checksums, report identifiers, baseline files, or differences. Automated QA browsers are excluded.

## Explicit limitations

- Static homepage source does not prove that a disclosure was visible at first interaction.
- A token or wording candidate does not prove an AI integration is active or legally in scope.
- The scanner does not download media and therefore does not verify Content Credentials or watermarks.
- No-change means only that this bounded normalization observed no difference.
- A checksum does not authenticate identity or establish chain of custody.
- The tool does not provide legal advice, a compliance score, certification, or an Article 50 conclusion.

## Primary references

- European Commission, Article 50 transparency guidelines: <https://digital-strategy.ec.europa.eu/en/library/guidelines-transparency-obligations-providers-and-deployers-ai-systems>
- EU AI Act Service Desk, Article 50: <https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50>
- European Commission, transparency rules quick facts: <https://digital-strategy.ec.europa.eu/en/factpages/quick-facts-transparency-rules-ai-systems>
- C2PA 2.4 AI disclosure: <https://spec.c2pa.org/specifications/specifications/2.4/specs/C2PA_Specification.html#_ai_disclosure>

H/M does not claim that this tool certifies Article 50 compliance or C2PA conformance.