Why it matters
Without AI, misattributed marks, undisclosed restorations, and reference errors arise, risking provenance disputes, compliance breaches, damaged consignments, and disputes over insurance and authentication.
From design to safety
Design
AI informs mark strategy and cataloging
The system reads reference databases, mark sketches, paste recipes, glaze catalogs, workshop notes, and archival provenance files to propose a consistent attribution framework and scalable catalog metadata aligned with cataloging standards.
- factory marks
- pastes
- glazes
- decoration
Build
Data capture for attribution framework
It ingests high-resolution macro and micro images, laser surface scans, porcelain tissue micrographs, reference literature, archival photos, and documented restorations to validate marks, detect forged references, and verify dating confidence.
- factory marks
- pastes
- glazes
- decoration
Operation
Ongoing attribution checks in sale cycle
The AI cross-checks new acquisitions against reference datasets, flags uncertain attributions, generates provenance notes for listings, and maintains a living change log tied to lots and consignor histories.
- factory marks
- glazes
- decoration
- reference marks
Maintenance
Periodic re-validation of marks
Periodic re-scans of marks and glaze layers, updated reference alignments, aging glaze analyses, and archival cross-checks ensure marks and restored surfaces remain accurately attributed and auditable for governance and insurance.
- factory marks
- pastes
- decoration
- reference marks
Safety
Compliance and documentation guardrails
Compliance checks enforce disclosures, reversibility notes, pedigree records, and provenance audits meet vetting standards, insurance requirements, legal disclosure, and ethical guidelines during consignments, vetting, and transfer of ownership.
- factory marks
- pastes
- glazes
- decoration
Part by part
- 1
factory marks
- Watch for
- Forgery risk on marks with multiple variants and post-1920 reproductions indicating misattribution, regional deviations.
- AI action
- Read mark registry, compare with known templates, trigger attribution confidence score and review flag.
- 2
pastes
- Watch for
- Inconsistent paste composition suggesting later restoration or misdated application within older layers of coating.
- AI action
- Analyze paste mineralogy against reference libraries and flag anomalies for expert review and documentation.
- 3
glazes
- Watch for
- Unmatched glaze layers or overglaze enamel indicating non-original surface work on older pieces.
- AI action
- Spectrographic glaze analysis with authenticity flags and dating hints for staff review and calibration.
- 4
decoration
- Watch for
- Decorative motifs that diverge from period style or workshop signatures in the record sets.
- AI action
- Pattern recognition against workshop catalogs, alert on deviations for vetting and archival reference.
- 5
forged marks
- Watch for
- Marks overlaid onto porcelain by later makers or counterfeiters in regional markets during production bursts.
- AI action
- Cross-reference with workshop inventories and material clues, raise alert for archival verification.
- 6
reference marks
- Watch for
- Reference marks that conflict with catalogued factory records or regional dating across collections.
- AI action
- Align with archival records, update provenance notes and dating confidence in the system log.
Worked example
Imperial-era mark verification on a famille-rose vase
Illustrative scenario with sample data.
A dealer examines a famille-rose vase and questions a 19th‑century mark that resembles a known imperial pattern, seeking decisive AI guidance.
Signal in
markCode: 123-Imperial, glaze: famille-rose, decoration: peony, provenance: EU, restorationHistory: 'repainted foot 2018', photos: 12, condition: good, dimensions: 28cm tall, weight: 1.2kg, owner: private
System out
Attribution note created: markCode validated against database; provenance flagged for further vetting; recommended refer to three reference specimens and schedule expert review; audit trail updated.
Human sign-off: Valuer approves the provenance note and assigns a conservator to verify the imperial mark against archival records; vetting committee signs final decision after review.
How to roll it out
Phase 1
Weeks 1 to 4
Baseline data ingestion, reference library setup, and initial mark-reading tests with extended reference sets.
Phase 2
Weeks 5 to 8
Expanded databases, cross-referencing with deaccession records, and first automated attribution reports fed to listings.
Phase 3
Weeks 9 to 12
Operational integration with listings, vetting checks, and staff training across departments.
Where people stay in charge
- All attribution decisions above a defined confidence threshold must be signed off by a conservator-restorer with accreditation.
- No attribution decisions involving endangered materials may proceed without ethics officer review.
- Any forged or high-risk marks require immediate escalation to the vetting committee for archival corroboration.
Questions owners and crews ask
How does AI help with provenance when attribution is ambiguous?
AI compares marks, pastes, and glaze data against curated reference sets, highlighting uncertainties for the human expert to confirm or deny with archival sources.
Can AI detect forged marks and non-original decoration?
Yes, it analyzes feature patterns, material signatures, and historical references to raise alerts when inconsistencies appear, prompting expert vetting.
What records are kept for each attribution?
Provenance notes, reference image sets, material analysis results, and vetting decisions are stored with audit timestamps for traceability.
Who signs off on final attribution reports?
A qualified conservator-restorer along with the valuer or designated vetting committee signs off after review.
Sources and further reading
- CINOA: Code of EthicsDealer ethics on provenance, due diligence, and disclosure for listed objects.
- GOV.UK: Money Laundering Supervision for Art Market ParticipantsAnti-money laundering registration and checks that art and antique dealers must complete.
- American Institute for Conservation: Collection CareCeramics care guidance covering glaze condition, handling, storage, and environmental risks for collections.
- V&A: ConservationMuseum conservation examples for ceramics, glazes, and surface examination methods used by curators.
- Getty Conservation InstituteMaterials analysis research on ceramic bodies, pastes, and composition testing methods.
- FBI: Art CrimeReporting routes and case examples of art and antique forgery, theft, and counterfeit objects.
- BADA: AdviceDealer guidance on identifying antiques, including marks, makers, and attribution checks.
- BADA: The BADA StandardDealer code of practice covering description accuracy, vetting, and buyer protection.
- American Institute for Conservation: Code of EthicsConservation ethics on reversibility, documentation, and guidelines for treatment practice.
- American Society of AppraisersPersonal property appraisal practice and valuation standards for antiques and collectibles.