Aqta Technologies proposes IETF standard for AI inference attestation receipts

Immediate adoption of this format would allow streaming providers to generate tamper-evident audit trails for AI-driven content moderation or automated licensing decisions. As platforms increasingly rely on large language models (LLMs) to handle B2B workflows, the ability to verify decisions offline becomes a critical compliance requirement. This standard bridges the gap between internal logs and independent legal or regulatory review by pinning evidence to a trusted public key. In the broader ecosystem, this move aligns with growing demands for transparency in automated systems without requiring full disclosure of proprietary model weights or datasets. Watch for IETF adoption of post-quantum signature suites in future drafts to address long-term audit retention risks.

The timing of Aqta Technologies' proposal coincides with the enforcement of Article 50 of the EU AI Act, which became applicable on August 2, 2026. Per official European Commission guidance from July 2026, providers of certain AI systems must now ensure outputs are marked in a machine-readable format to allow identification as artificially generated or manipulated content. Non-compliance with these transparency obligations carries potential fines of up to €15 million or 3% of total global turnover. The IETF draft aims to provide a standardized technical mechanism for meeting these rigorous documentation requirements across different jurisdictions. Beyond simple transparency, the broader standards landscape is rapidly fragmenting around AI provenance. Per IETF LLM standards proposal records from July 2026, related efforts such as the Attested Inference Receipt (AIR) profile and the Composing Application-Layer Action Evidence with Remote Attestation Procedures (AEP) draft are also under review. While Aqta’s proposal focuses on lightweight application-layer evidence, the AIR draft specifically profiles COSE and CWT for use within confidential computing environments like Trusted Execution Environments (TEEs). The competition between these formats will likely determine how streaming platforms integrate compliance tools into their existing delivery stacks, particularly for high-stakes automated decisions.