AEP
Attestation& Evidence
Exchange Protocol
Immutable cryptographic attestation and verifiable evidence infrastructure for AI agents and AI creation.
An open evidence protocol for AI creation and agent behavior — turning every generation into an independently verifiable cryptographic evidence chain.
The Trust Crisis of the AI Era
Creation has grown exponentially; the ability to prove authorship has stood still.
Creations are copied, provenance is unprovable
AI images, video and text are created, shared and commercialized at scale, yet it is hard to prove who made them, with which model, under what parameters, when, and without plagiarizing training data.
It freezes the entire creative process into a cryptographically verifiable evidence chain that any party can verify independently, outside the original environment.
Attestation tools don't know AI; AI tools don't attest
Blockchain notarization stores hashes without understanding semantics; AI tools produce content without keeping process records. The two communities talk past each other, with no translation layer between them.
AEP is that translation layer between cryptographic attestation and AI creation, providing a complete, unified evidence system for the creative process.
Compliance and courts are waiting for a technical answer
AI content labeling is now a mandatory national standard; courts face three hurdles in AI cases—fact-finding, legal application, and liability attribution—while existing evidence only proves "a file existed at some time."
It turns the creative process into structured evidence admissible to courts and regulators, underpinning compliance obligations and legal proof.
Six Core Use Cases
Providing tamper-proof objective evidence for creative iterations, generative assets, and mission-critical agent actions
Papers, Patents & Software Copyright Protection
End-to-end evidence retention for research derivation, experimental telemetry, prompt evolutions, and code commits, providing verifiable evidence of priority and attribution.
Multimedia Generative Assets (Audio / Video / Image)
Full-chain provenance for generative art, synthesized audio, and video editing/remixing to prevent copyright laundering and illicit derivative works.
AI Coding & Code Supply Chain Anti-Poisoning
Trace code diffs, dependency additions, and runtime environment probes by agents like Cursor or OpenClaw, providing evidence for post-incident attribution and audit.
Judicial Evidence & Dispute Arbitration
Export structured .aep evidence packages that support submissions to internet courts, notaries, and arbitration panels, easing fact-finding.
Mission-Critical Agent Actions & Automated Approvals
Solidify non-repudiable execution receipts, prompts, and complete contexts when LLMs execute autonomous operations, procurement, payment transfers, or policy approvals.
AIGC Copyright & Originality Proof
Fully capture prompts, parameters, multi-round curation, and human post-processing to provide verifiable evidence of human intellectual contribution when asserting copyright over AI-generated content.
Who owns the copyright to AI-generated content?
Purely AI-generated work with no human involvement is not protected by copyright; AEP turns "human intellectual contribution" into verifiable process evidence.
Copyright law and judicial practice establish that a work must reflect a natural person's original intellectual input. Issuing a single command to an AI is merely advisory assistance, not authorship.
Content generated solely from simple, generic instructions, without any personalized intervention or selection and wholly produced by AI at random, lacks originality, does not constitute a work, and is not protected by copyright law.
The four yardsticks courts use to assess originality
Prompt design
Personalized, specific choices about theme, style, composition, wording, and scene detail, rather than a simple generic instruction.
ai.input · full prompt & evolution Parameter & model control
Tuning core parameters such as composition ratio, clarity, style weight, iteration steps, and random seed, showing control over the result.
ai.generation · params / modelHash Multi-round curation
Comparing, rejecting, and retaining multiple candidate outputs based on creative intent and personal aesthetic judgment.
outputs[] · selected: true Post-generation human editing
Materially modifying, editing, polishing, or restructuring the AI output into the final expression.
ai.postprocess · event chain An independently verifiable originality / copyright signal
The four-factor process evidence is digitally signed, timestamped, and on-chain anchored, then sealed into a standard .aep package exportable to courts, notaries, or copyright registries.
Standards & legal sources
AEP provides objective, neutral cryptographic process evidence and does not substitute for a court's final determination of originality in any given case.
View the AI Profile spec