CONTENT PROVENANCE
AI Content Transparency Check
Check an image for Content Credentials and supported C2PA provenance information. Analysis runs locally in your browser.
Check Content Credentials in an image
The AI Content Transparency Check is a free Content Credentials checker for images. It examines the selected file for supported C2PA provenance information and presents the information available in the file. The analysis runs locally in your browser. Your image is not uploaded to Alterlayer for analysis.
Use the checker when you want to know whether an image carries Content Credentials, whether supported provenance information can be read and what that information says about the image’s recorded origin or editing history. A result should be treated as one source of evidence, not as a final decision about whether the image, its subject or the context in which it appears is trustworthy.
For the most useful result, select the original image file whenever possible. Screenshots, social-media downloads, resized exports and files processed by unsupported services may no longer contain the provenance information that was attached to an earlier version.
What are Content Credentials?
Content Credentials are provenance information associated with digital content. They can record statements about how a file was created, which tools were involved and what changes were made during a supported workflow. They are designed to give people additional context about a digital asset rather than requiring them to judge the asset only from its visible appearance.
A useful comparison is a nutrition label for digital content. The label does not decide whether the content is good, accurate or appropriate. It provides information that can help a person understand the content’s recorded history and make a better-informed decision.
Content Credentials can be attached by a supported camera, software application, content platform or generative AI system. The available information depends on the product that created the credential, the assertions it included and what happened to the file afterwards. One image may contain detailed provenance information, while another may contain only limited information or no readable credential.
Unlike ordinary descriptive metadata, C2PA Content Credentials can include cryptographically signed claims that are bound to a particular digital asset. This allows compatible verification software to check the relationship between the claim and the file and to identify certain integrity or validation problems. It does not make every statement inside a credential automatically true, and it does not establish that the visible subject of the image represents a real event.
What is C2PA?
C2PA stands for the Coalition for Content Provenance and Authenticity. The coalition develops an open technical standard for recording and verifying provenance information for digital content. The standard is used by participating technology, media, camera, creative-software and AI organizations to support interoperable Content Credentials.
A C2PA credential can contain assertions about an asset, a claim that groups those assertions, a digital signature and a binding between the claim and the corresponding content. Compatible validation software can examine these elements and report whether the supported manifest is well formed, whether its signature and bindings can be validated and which provenance statements are available.
C2PA is not a database that declares which images are true. It provides a technical structure for provenance and integrity signals. The usefulness of those signals still depends on who or what issued them, which information was recorded, whether the credential remains available and how the image is being used.
The standard can support more transparent media workflows because provenance information can travel with a digital asset or be associated with it through supported mechanisms. Adoption is still developing, however, and not every camera, editing application, AI system, publishing platform or social network preserves or displays Content Credentials.
How this C2PA checker works
Select an image through the tool above. The checker uses the existing browser-based C2PA implementation to inspect the file for supported Content Credentials and provenance data. When supported information is available, the result presents the information returned by the verifier. When no supported credential can be found or validated, the result must be read with the limitations explained below.
- Select the image you want to inspect.
- The browser reads the file using the existing local verification implementation.
- The tool checks for supported C2PA Content Credentials and available provenance information.
- Review the result and its recorded evidence before making any decision about the image.
The image itself is not sent to Alterlayer for this analysis. This local-processing approach is particularly useful for confidential, unpublished or commercially sensitive images that should not be transferred to an additional analysis service.
Local processing does not change the evidential limits of Content Credentials. It protects the selected file from being uploaded to Alterlayer, but it does not create missing provenance information, recover every credential removed by another platform or verify facts that were never recorded in the file.
How to interpret a Content Credentials result
Content Credentials are found
When supported Content Credentials are found, review the available issuer, creation, tool, action, signature, validation and provenance information presented by the checker. The exact fields depend on what was included in the credential and what the current verifier can read. The presence of a credential means that provenance information is available for examination. It does not, by itself, prove that the image is factually accurate or being shown in its original context.
No supported Content Credentials are found
A “no credentials” result means that the checker did not find supported Content Credentials in the selected file. It does not mean that the image is authentic, fake, human-created or AI-generated. The image may never have carried a credential, or an earlier credential may have been removed when the file was copied, compressed, resized, exported, screenshotted or processed by another service.
Credentials cannot be fully validated
If provenance information is present but cannot be fully validated, do not treat the credential as verified evidence. The file may contain unsupported, incomplete, damaged, altered or otherwise unverifiable information. Preserve the original file and review the detailed result before relying on the credential in a publishing, governance or investigation workflow.
Content Credentials are evidence, not a truth verdict
Content Credentials help answer questions about recorded provenance: where a supported file came from, which tools were involved, which actions were declared and whether the associated technical claims can be validated. They do not independently determine whether the people, objects, statements or events shown in an image are real.
A photograph of a staged event can carry valid Content Credentials. An accurately documented image can be presented with a misleading caption. A genuine photograph may have no Content Credentials at all. A synthetic image may lose its credential when it is copied through an unsupported workflow. Provenance evidence should therefore be reviewed together with the source, publication context, editorial process and any other available evidence.
For high-impact decisions, preserve the original file, document where it came from, review the available provenance information and escalate uncertainty to an appropriate human reviewer. Do not use a single automated result as the sole basis for a legal, employment, financial, safety, journalistic or reputational decision.
Content Credentials checker vs AI image detector
A Content Credentials checker and an AI image detector answer different questions. A Content Credentials checker looks for recorded provenance information and supported cryptographic evidence associated with the file. A typical AI image detector estimates whether visual or statistical patterns resemble content produced by an AI model.
The Alterlayer tool performs the first function. It checks supported Content Credentials and C2PA provenance information. It does not assign an AI probability score, inspect visual anomalies or claim to recognize every image produced by Midjourney, Stable Diffusion, ChatGPT, Gemini, Firefly or another generative system.
When a supported credential explicitly records that an AI system or trained algorithm was involved, that information may help establish the image’s recorded origin. When no such credential is available, the checker cannot conclude that AI was not involved. This distinction avoids presenting absence of evidence as evidence of absence.
For organizations establishing wider controls around generated content, provenance evidence can form part of a broader AI governance and evidence process.
Why image provenance matters
Images move rapidly between creation tools, internal teams, agencies, publishing systems, messaging applications and social platforms. During that journey, it can become difficult to determine where a file originated, which tools were involved and whether the version being reviewed is still connected to its original provenance information.
Content Credentials provide one technical method for carrying recorded provenance with digital content. When credentials remain available, reviewers may be able to inspect information about the tool or organization that issued the credential, declared creation or editing actions and the integrity of the associated manifest.
This information can support more disciplined decisions, but it must remain connected to organizational context. A technical provenance signal does not identify who approved publication, whether the image is appropriate for its intended audience, whether the caption is accurate or who remains accountable for the final use of the content.
Organizations therefore need both technical signals and operating controls. Provenance can help explain what information is available about a file. Governance establishes who reviews that information, what additional verification is required, who can approve use and what evidence must be retained.
Who can use an image provenance checker?
Communications and marketing teams
Communications teams can inspect original or supplied images before publication and record whether supported provenance information was available. The result can inform an editorial review, but it should not replace confirmation of the image source, usage rights, publication context and accountable approval.
Newsrooms and publishers
Editors and journalists can use a C2PA checker as one step in a wider source-verification process. A credential may provide useful information about the file’s recorded history. It cannot independently confirm that an event occurred, that a caption is correct or that the person supplying the image is reliable.
Legal, risk and compliance teams
Legal, risk and compliance functions can use available provenance information when reviewing how synthetic or edited media is created, received and published. The result may support documentation and escalation, but it does not constitute a legal conclusion or demonstrate compliance with a particular regulation.
Procurement and third-party management
Organizations receiving images from agencies, suppliers or technology providers can include provenance review in their acceptance process. They may ask which tools created the content, whether Content Credentials are expected, what transformations occurred and which party is responsible for final approval.
AI governance teams
AI governance teams can connect content-level provenance with wider records about the AI system, model or service involved. Relevant organizational records may include ownership, purpose, permitted use, approval conditions, human review, publication responsibility and supporting evidence.
A practical image-review workflow
A Content Credentials result is most useful when it is part of a repeatable review process. The following workflow separates the technical check from the human and organizational decisions that follow it.
- Preserve the original image file whenever possible.
- Record where the image came from and who supplied it.
- Use the Content Credentials checker to inspect available C2PA provenance information.
- Review the issuer, declared actions, tools, validation status and other available details.
- Compare the result with the image’s visible content, caption, source and intended use.
- Escalate missing, inconsistent or unverifiable information when the decision could have a material impact.
- Record who reviewed the image, who approved its use and which evidence supported the decision.
The appropriate depth of review depends on the context. A decorative internal image may require little additional verification. An image used in public reporting, regulated communications, a legal matter, a financial decision or a reputationally sensitive publication may require stronger source confirmation and documented human approval.
Why Content Credentials may disappear
Content Credentials do not necessarily survive every transformation or distribution channel. Some applications and platforms preserve supported provenance data, while others may remove metadata when they compress, resize, convert or re-export a file.
A screenshot usually creates a new image rather than preserving the original file structure. Copying an image through a messaging application, social network, document editor or content-management system may also remove or alter embedded information. An image that once carried Content Credentials may therefore reach a reviewer without them.
Some ecosystems use additional mechanisms such as external credential storage, digital fingerprinting or invisible watermarking to help reconnect transformed content with provenance information. Support varies between providers, and this checker must not be described as recovering every removed credential or detecting every proprietary watermark.
When provenance is important, retain the original file, avoid unnecessary transformations and document the transfer path. A result produced from a downloaded or modified copy may not represent the information that was available in the original asset.
Content Credentials and the EU AI Act
Article 50 of the EU AI Act applies from 2 August 2026 and includes transparency obligations for certain AI systems and AI-generated or manipulated content. In particular, providers of systems generating synthetic audio, image, video or text content must ensure that relevant outputs are marked in a machine-readable format and detectable as artificially generated or manipulated, subject to the scope and exceptions established by the legislation and related guidance.
Content Credentials and C2PA can contribute to machine-readable provenance and transparency, but using C2PA does not automatically establish compliance with the EU AI Act. The applicable obligation depends on the organization’s role, the AI system, the content, the way it is used and the relevant legal conditions.
The Alterlayer checker does not perform an EU AI Act compliance assessment. It checks the selected image for supported Content Credentials and C2PA provenance information. Organizations remain responsible for determining which transparency, marking, disclosure, governance and evidence requirements apply to their activities.
Private image checking in your browser
The AI Content Transparency Check performs its existing supported analysis locally in the browser. The selected image is not uploaded to Alterlayer for analysis, and the tool does not create a public image record or a crawlable result page.
Local processing reduces unnecessary disclosure of confidential or unpublished image content. It can be useful when teams need to inspect a sensitive file without transferring the image to another analysis service.
Browser-side processing does not make the result anonymous in every possible sense and does not override the operation of the website’s existing essential infrastructure or approved anonymous analytics. However, the selected image and its visual content must not be included in analytics events, logs, contact requests or external notifications.
Do not use the tool for material that you are not authorized to access or process. Users remain responsible for the files they select and for the decisions they make after reviewing the result.
Frequently asked questions
What is a Content Credentials checker?
A Content Credentials checker examines a digital file for supported provenance information based on the C2PA standard. It may display available information about the recorded origin, tools, actions, claims, signatures and validation status associated with the file.
How can I check whether an image has Content Credentials?
Select the image in the tool above and review the result produced in your browser. Use the original file whenever possible because screenshots, compressed downloads and re-exported copies may no longer contain the provenance information attached to an earlier version.
Is this an AI image detector?
No. The tool checks supported Content Credentials and C2PA provenance information. It does not analyze visual patterns or produce a probability that an image was generated by AI. If a supported credential records AI involvement, that information may appear in the available provenance data.
Does no Content Credentials found mean the image is real?
No. It means that the checker did not find supported Content Credentials in the selected file. The image may never have contained a credential, or the credential may have been removed or become unavailable during copying, compression, resizing, screenshotting or export.
Do valid Content Credentials prove that an image is true?
No. Valid credentials can provide verifiable provenance information associated with a file. They do not prove that the depicted event occurred, that every statement is factually correct, that the caption is accurate or that the image is being presented in the proper context.
Can Content Credentials be removed from an image?
Yes. Some tools and platforms may remove embedded metadata when they process an image. Supported ecosystems may use additional methods to reconnect a transformed file with externally stored credentials, but recovery is not guaranteed across every workflow.
Is my image uploaded to Alterlayer?
No. The existing supported analysis runs locally in your browser. The selected image is not uploaded to Alterlayer for analysis, is not stored as a public record and does not receive a public result URL.
What should I do if the result is inconclusive?
Preserve the original file, document its source, review the detailed result and seek additional evidence. For material decisions, ask an appropriate human reviewer to assess the image, its provenance, its context and the consequences of using it.
Can this tool demonstrate EU AI Act compliance?
No. The checker provides information about supported Content Credentials and C2PA provenance in an image. It does not determine which EU AI Act obligations apply to an organization or demonstrate that all applicable transparency, governance and evidence requirements have been satisfied.
