Spiritual screening for AI: Russia introduces mandatory checks of models for traditional values
The Russian artificial intelligence market is on the verge of tectonic changes. Since July 26, a framework law on AI has come into force, which is now being fleshed out with specifics. The key innovation, actively discussed in the professional community on the platform of the ANO "Digital Economy," is the introduction of mandatory certification for foundational models claiming the status of "sovereign" or "national." This is not just about a technical audit, but about checking compliance with legislation and, more interestingly, with traditional spiritual and moral values.
How will the new system work?
The mechanics are as follows. A developer wishing to gain access to state support and preferences in public procurement will need to apply to an accredited testing laboratory. A package of documents is submitted there: an application, the model architecture, and a description of request filtering mechanisms. The laboratory, in turn, will conduct a series of tests using special benchmarks, checking not only the correctness of responses but also resistance to prompt injections—attempts to hack AI with malicious instructions hidden in input data.
An important nuance: a unified list of values that will form the basis of the check does not yet exist. Instead, it has been decided to develop sets of test tasks jointly with businesses. The conclusion itself will be issued by a special digitalization commission, and it is a positive verdict that will become the pass to government contracts. At the same time, they plan to check not everyone indiscriminately, but only those models that claim a high status. The first areas of implementation will be education and public services.
Who will issue the verdict?
The organizational scheme is also interesting. Instead of a single monopoly center, several laboratories that have passed verification by a government body will be able to conduct the expertise. A two-stage system is envisioned: first, the developer independently tests the model using a general risk-based methodology, and then an accredited laboratory evaluates these findings and may selectively verify resilience. The final decision on status will remain with the Ministry of Digital Development, while security issues for government systems will be overseen by the FSB and FSTEC.
Well-founded concerns are already being voiced in the industry. Alexey Borshchov, business architect of the AI ecosystem "Avandok," rightly notes that categories such as civic consciousness or historical memory are extremely difficult to translate into a technical specification. And Dmitry Galantsev, managing partner of the law firm "Propositum," warns of the risk of "subjectivity" in the criteria, which could lead to selective application. He proposes introducing an accrediting body, several competing laboratories, and an appeals board, as well as a public registry of conclusions and certificate validity periods.
Analyst's comment: This is a landmark step that moves AI regulation from a purely technical plane into an ideological one. For the market, this is a double challenge: on the one hand, a clear barrier to entry into the public sector appears; on the other, a high degree of uncertainty in the interpretation of "spirituality" could slow down innovation. The success of the initiative will directly depend on how transparent and reproducible the assessment procedure becomes. Otherwise, we risk getting not a security tool, but a mechanism for manual market management.