Disclara product and technical updates

This is a manual report workflow, not document upload, Office automation or independent C2PA validation. Final-file reports must match the saved asset hash when recorded; otherwise byte binding stays explicitly unestablished. Receipt hashes are unsigned, and no report grants publication clearance. Existing checker decisions and paid-launch gates remain unchanged.

No rule or source was reinterpreted or reverified by this catalog release. Scope matches are not legal decisions; provider marking, client policy, provenance and observed enforcement remain explicit gaps or full-checker dependencies. No invented workflow experiment or customer benchmark is included. Existing evaluator archives and saved decisions are unchanged.

Disclara has no Claude detector or automatic platform monitoring. Non-detection is not human-origin proof. Manual observations are limited to a surface, region, time and variant; metadata from a downloaded copy is not exact served-byte evidence. Observations never override the original publication/review gate. Paid checkout remains disabled.

This is not an activated paid offer. OVH email and Stripe adapters require configuration/approval. Legacy decisions without a bound evaluator remain replay-unsupported; registry metadata is not a primary-source body. Human approvals do not override checker findings, and hashes are not signatures or trusted timestamps.

Facts and attribute comparisons are caller-supplied. MP4/audio inspection, automated image/PDP comparisons, feed connectors and platform-label certainty are not available. Missing scope or evidence requires review; no TikTok Shop rule is inferred for other markets or ordinary posts.

These are review signals from caller-supplied facts. They do not prove a legal transfer, authenticate event order, approve the final asset or create a signed SOW/MSA. The schedule requires acceptance and legal review outside Disclara.

A prompt, approval or platform account is not treated as conclusive proof of legal deployer status. The matrix is an unsigned, caller-supplied evidence record and does not generate a binding contract or independently verify a party's control.

This is a marketing and product-availability update, not a subscription launch. Waitlist, trial and checkout remain disabled; no customer data or paid entitlements are created by this page.

This changes presentation and input safety, not the underlying legal obligation or the availability of paid workspaces. The example is not an asset inspection, a completed check or legal approval.

REPRODUCED means the newly selected bytes match this deterministic export method and record; it does not authenticate who made the record, whether a browser download succeeded, prior local recheck, policy approval, disclosure, privacy completeness or publication. Invalid or unsupported edits remain inconsistent rather than silently accepted.

A native-disclosure declaration is no longer left without platform context in new downloads. The selected destination, client policy, approval and disclosure are still unsigned caller claims, not authenticated platform events. Local Exif export records now use version 2 because their nested comparison capture changed.

This limits page-thread blocking from malformed captures or expensive local scans. Files are not uploaded; a browser worker is not an operating-system sandbox and does not authenticate the unsigned capture or publication evidence.

REPRODUCED INCONCLUSIVE confirms deterministic replay of selected bytes only. It does not establish absence, preservation or safe removal of provenance; the user must obtain readable files and recheck. Approval, policy, disclosure and publication remain unauthenticated.

REPRODUCED means the newly selected bytes match the capture's deterministic comparison. The JSON remains unsigned: the replay cannot authenticate the author, past local-copy recheck, approvals, client policy, disclosure, download, upload, served asset or timestamp. Other evidence-envelope formats remain in their separate verifier.

An earlier asynchronous file read can no longer overwrite a newer source or candidate selection. The capture remains unsigned and does not prove that a download, upload or publication occurred. Existing standalone mutation decision v2 is nested inside capture v1; consumers of downloaded JSON should account for that format change.

Matching selected bytes do not prove that a download occurred or identify a filesystem path. The unsigned record does not establish permission, disclosure, upload or the served platform variant; files remain in the browser.

The export blocks files with detected C2PA, XMP or IPTC, unclassified JPEG APP/comment segments or trailing data, multiple Exif segments, MakerNotes, nested/thumbnail IFDs and malformed or overlapping value storage. It does not remove artist/owner identity or prove that all private data is gone. No signed credential is updated; publication, client-policy and disclosure review remain separate.

Malformed Exif is inconclusive; any changed file still requires review, and unchanged C2PA presence does not establish integrity after export. MakerNotes, XMP privacy fields and other metadata are outside this scan. No selective redaction, signed receipt or publication observation is provided.

Lost provenance or changed bytes with C2PA trigger review; a declared client prohibition blocks the changed candidate. EXIF is located but private fields are not decoded, no metadata is removed, and the unsigned record does not prove actual publication, signer trust or legal sufficiency. Safe selective redaction remains unavailable.

Unbound and mismatched final-asset reports remain UNVERIFIED and trigger review. Original-capture and component claims keep their own narrower scope; a hash match only compares caller-supplied bytes and does not authenticate a report or establish signer trust. Evidence format v17 records the binding state.

IPTC/XMP remains declarative, incompatible scheme/claim pairs remain unsupported, and a vulnerable C2PA validator cannot yield a reported-valid interpretation. Complete reports are still caller-supplied, not Disclara signature checks or proof of scene truth. Evidence format v16 records the report gate and its boundary per claim.

The tool distinguishes intact embedded credential integrity from signer trust and platform-label interpretation. It has a 10 MB file limit and 12-second deadline; it does not fetch external manifests, check revocation or authenticate rights. A browser worker is not an OS-level parser sandbox, and the disclosure checker still accepts caller-supplied verifier evidence rather than directly parsing assets.

A reported C2PA result from an affected validator cannot be used as dependable positive provenance evidence. The gate checks a tracked advisory and the completeness of the supplied record; Disclara does not yet parse assets, verify C2PA signatures or authenticate the caller's attestation. Automated sandboxed verifier execution remains planned.

Teams can keep the media binary in their existing DAM while downloading a replayable Disclara decision-provenance envelope. The builder does not retrieve or store assets, authenticate external references, validate C2PA signatures or operate a live Bynder, Fotoware, Adobe, API or webhook connector. Those integrations and a retained Evidence Graph remain planned.

Teams can see the metadata container, byte location, parser status and Digital Source Type behind a label-risk estimate. Presence still does not mean that a signature is valid, and compressed metadata that cannot be decoded remains explicitly unresolved.

Teams can identify when technically valid provenance may create an overbroad platform label, while California video and audio ads are evaluated using performer type, voice type, real-person representation and commercial role rather than a generic AI boolean.

Marketing teams can record and interpret more precise AI-use provenance without treating a valid credential as proof of truth, authorship or legal compliance.