The Archive Becomes Infrastructure — But It Doesn’t Stop Being Memory
China’s new Five-Year Plan for archives reads like a blueprint for digital government. The more consequential question is who gets to write the standards underneath it.
Most national archival strategies are predictable documents. They promise more digitization, better access, improved storage, and, increasingly, some use of artificial intelligence. The technology changes. The institutional self-image usually does not. The archive remains a custodian of memory, now equipped with newer tools.
China’s National Archival Development “15th Five-Year” Plan initially appeared to break that pattern. The roadmap for 2026 through 2030 speaks not only to collections and preservation but also to artificial intelligence, high-quality datasets, trusted authentication, cybersecurity grading, commercial cryptography assessment, backup centers, and a national query-and-use platform. It was issued jointly by the General Office of the Communist Party Central Committee and the General Office of the State Council. That makes it more than a statement of archival ambition. It gives the plan authority and reach beyond a strategy issued by the archives administration alone. In the Chinese planning system, it signals that other agencies are expected to take the plan seriously and incorporate it into implementation.
My first reading, however, pushed the infrastructure argument too far.
China is not replacing the archive as a memory institution. Much of the plan remains grounded in familiar archival responsibilities: building collections, opening records, supporting exhibitions and historical research, protecting holdings from physical threats, developing professional talent, and participating in international cultural-memory programs through the ICA and UNESCO’s Memory of the World. The traditional formulation is still there in the opening lines: archives should preserve history, inform governance, and educate people.
What the plan adds is a second institutional identity. The archive must continue to preserve memory while also becoming part of the operational infrastructure of digital government. Nothing substantial has been removed from the older mandate. A layer of data governance, trusted systems, artificial intelligence, and cyber resilience has simply been placed on top of it. The story is not replacement but accretion — archives asked to function simultaneously as memory institutions and data infrastructure. National archives elsewhere use different legal and institutional models, but they are drifting toward the same double burden, and I will spend the rest of this piece on the consequences rather than the claim.
Machinery I recognize
After years of watching electronic records policy in the United States shift from repositories to systems, I recognize the architecture being assembled here. The vocabulary is Chinese, and the institutional arrangements are different, but the underlying problem is one I have seen NARA wrestle with for a decade: how do you exercise archival control when the record no longer arrives as a bounded object?
The plan’s answers are specific. It calls for “AI + archives” (人工智能+档案), for archival data governance (档案数据治理), for building high-quality datasets, and for cautious exploration of large models — and, in the same breath, for full-lifecycle data-security management across AI design, training, deployment, and use, and for an “algorithm risk monitoring and early-warning mechanism” (算法风险监测和预警机制). AI is treated as an operating layer and as a governed hazard at once. That pairing is more disciplined than most Western policy language I have read, and it opens a question the AI-governance literature almost never asks: when does an AI interaction create a record? I will come back to it.
Two other commitments matter. The plan promotes the filing and transfer of records “in electronic single-copy form” (单套归档并移交) — the quiet end of the hybrid paper-and-digital compromise. I want to be precise about the size of that step, because it is easy to oversell. The verb is promote, not mandate, and China already issued a national standard on single-copy management, DA/T 92, back in 2022. This is the scaling of an existing trajectory, not a 2026 pivot. The direction, though, is not in doubt. It belongs to a broader international transition toward authoritative digital records, although the legal meaning of the authoritative copy, the transfer model, and the role of the national archive differ considerably across jurisdictions. NARA’s long transition toward electronic recordkeeping and electronic transfer is the case I know best; Australia’s Digital Continuity work is another, reached by a different route.
The plan also widens what counts as an archival object: public data, business data, scientific data, electronic invoices, electronic licenses. The custodial unit was the record, then the electronic record, and it is increasingly extending toward the dataset. Whether archival science has a theory of the dataset comparable to its theory of the record is, to me, an open and slightly uncomfortable question. The literature contains important pieces of one, but I do not yet see a consolidated theory of the dataset with anything like the disciplinary standing of the archival record.
Looking Beyond the Horizon
The MetaArchivist has always been less interested in today’s headline than tomorrow’s architecture.
Most discussions about archives begin after change becomes obvious. I try to identify the standards, policy decisions, and technical designs that quietly reshape the profession years before most people notice. Horizon scanning is not simply watching the future arrive. It is recognizing the machinery that is already being assembled beyond the visible horizon.
The analysis that follows is where we move from the policy document itself into the standards ecosystem surrounding it: the committees writing the rules, the international connections, the implications for ISO, and what these developments may mean for archivists everywhere.
If you find value in this kind of deep archival intelligence, please consider becoming a paid subscriber. Your support makes it possible to continue the research, standards tracking, and primary-source investigations that turn obscure technical developments into practical insight for the archival profession.
