Products Lemma APIProof issuance & verification platformTrust402Delegate to agents, and transactSealSign-in for the agent era — no keys handed over
Use cases Manufacturing & Critical InfraInspection Record AssuranceFinance & FinTechCounterparty Record VerificationPublic SectorCertificate-less ProceduresHealthcareQualified Worker AssuranceProcurement & Supply ChainSupplier Credential MonitoringMedia & ContentContent AuthenticityService & RetailCross-group IdentityAI Adoption (cross-industry)AI Run GovernanceDevelopers & Agent OpsAgent Authority Control ▸ Browse the use-case index
Pricing
Resources Critical BriefThe frontier of AI × trustBlogThinking and implementation notesDocumentationAPI & specsVerification CenterReal verification & issuance countsAbout usFRAME00, Inc.ContactSales & press inquiriesGlossaryDefinitionsFAQFrequently asked questions
Dashboard ↗ JA
Glossary · Threat Model & AI Attack

Excessive Data Exposure

API の過剰なデータ返却

A design flaw where an API includes more fields in its response than the app or screen needs. The screen may show only a name, but if the response contains an address and phone number, anyone looking at the traffic can collect them.

Definition

The OWASP API Security Top 10 listed it as "Excessive Data Exposure" in its 2019 edition; the 2023 edition folded it into "Broken Object Property Level Authorization," which covers reading and writing fields a client isn't allowed to. It tends to happen when filtering is left to the client and the server returns database rows as they are.

Macnica's tally (published October 7, 2026, in Japanese) lists "APIs that return more information than necessary" among the weaknesses attackers targeted, and reports cases where API keys extracted from smartphone apps were used to call APIs in a way that looks like legitimate access. In 9 of the 81 cases disclosed since July 2026, data retrieval through APIs or databases was confirmed. The article recommends checking whether APIs return more data than needed and whether guest access returns unnecessary results.

Lemma implementation

Excessive data exposure comes from designs that hand over the raw data in order to establish a needed fact. Lemma answers a query with a proof that a condition holds, not with the raw data. To "is this member 20 or older?" it returns not a birth date but a zero-knowledge proof of that fact.

Which fields may be returned is fixed up front in the schema and the selective disclosure settings. For queries routed through Lemma, undisclosed attributes have no path into the response at all, so nothing depends on client-side filtering.

Which client received which attribute, and when, is kept in the audit trail. Whether reads wider than expected keep happening can be checked afterwards.

Get started

Return the needed fact, not the raw data.