Data freely usable, modifiable, and shareable by anyone, subject at most to attribution.
For media and cultural organizations, open data practice runs through licences: a dataset or collection is open when released under terms such as CC BY or CC0 that grant reuse and modification rights, with attribution the enforceable residue. The live operational question is scraping: whether works that are merely publicly accessible online count as open for AI training. Rights holders operationalize openness strictly by licence terms and text-and-data-mining reservations, while AI developers have often operationalized it as de facto public availability.
In practice: Distinguish licence-granted openness from mere public accessibility before reusing a dataset or collection; check attribution duties and any text-and-data-mining opt-out reservations first.
OmniGloss seed synthesis, 2026 (machine-drafted, pending expert validation)
In banking and payments, open data denotes regulated access regimes — open banking and open finance — in which account and product data held by institutions must be exposed through standardized APIs to authorized third parties when the customer consents. What is 'open' is the interface and the right of entry for licensed participants, not the data: the datasets remain personal, confidential, and tightly scoped by consent and authorization. Operational tests are API conformance, third-party accreditation, consent management, and liability allocation for misuse.
In practice: Map which account data must be exposed via open-banking APIs, verify third-party authorization and the scope of customer consent, and assign liability for data misuse across the access chain.
OmniGloss seed synthesis, 2026 (machine-drafted, pending expert validation)
In health research practice, 'open' data almost never means data anyone may freely download: patient-level data are shareable only after de-identification and through controlled-access arrangements — data access committees, data use agreements, secure environments. Researchers operationalize openness as 'as open as possible, as closed as necessary': registered access, documented approval criteria, and published metadata about what exists, while record-level data stay behind governance. Calling a health dataset open signals discoverability and a defined route to access, not unrestricted reuse.
In practice: Determine the correct openness tier for a health dataset — public aggregate, registered access, or committee-approved use — and document the de-identification and approval route for each tier.
OmniGloss seed synthesis, 2026 (machine-drafted, pending expert validation)
In government information management, open data is operationalized as the proactive release of public-sector datasets in machine-readable formats under a licence permitting anyone, including commercial actors, to use, modify, and redistribute them, subject at most to attribution. The tests a publication officer applies are concrete: is the licence on the approved open-licence list, is the format machine-readable rather than a PDF scan, is DCAT-conformant metadata attached, has personal-data and statistical-disclosure screening cleared release, and is any charge limited to marginal cost. Publication is the default position and refusal the exception, to be justified under the applicable access-to-information regime.
In practice: Assess a dataset for release: verify licence openness, machine-readability, metadata conformance, marginal-cost pricing, and personal-data screening, and justify any refusal to publish under the applicable access regime.
OmniGloss seed synthesis, 2026 (machine-drafted, pending expert validation)
In official-statistics institutes, open data is operationalized as dissemination discipline: statistical outputs are released to all users simultaneously, free of charge, on a pre-announced release calendar, with methodological metadata and quality reporting attached. 'Open' here is measured by impartiality and equal access — no privileged pre-release to ministers or markets — and by accessibility of formats, not by absence of confidentiality controls: underlying microdata remain protected, and only aggregates or safe files are opened.
In practice: Plan a statistical release so that all users receive it simultaneously under a published calendar, with metadata and quality documentation attached, while confidential microdata remain protected.
OmniGloss seed synthesis, 2026 (machine-drafted, pending expert validation)
The sectors draw the boundary of 'open' incompatibly. Public administration operationalizes open data as a universal reuse right: anyone, any purpose, open licence, at most attribution — access controls disqualify the label. Health research and finance attach 'open' to governed access arrangements — controlled-access repositories and consented, authorization-gated APIs — where data are neither public nor freely reusable, and openness means a discoverable, rule-bound route in. Both usages are institutionally entrenched, one in open-government law and the Open Definition, the other in research-governance and open-banking regimes, so the same word licenses very different expectations about who may obtain and reuse the data.