Around Yandex's "Alice AI," passions are running high: the market and users have been discussing the likelihood of granting the voice assistant expanded powers, up to and including access to banking apps and sensitive data of Russians. However, as my own data and analysis of the situation show, these concerns have no real basis.
Boris Ryutin, who is responsible for the security of "Alice AI" and Yandex's self-driving transport, gave a clear and unambiguous answer: the assistant has no "unlimited" access, and never will. He emphasized that Alice's capabilities on any device—smartphone, tablet, or smart speaker—are always strictly limited by the operating system's security mechanisms. The assistant can only work with those functions and data for which the user has given explicit and informed permission.
This is an important statement that shatters speculation that Alice would supposedly receive carte blanche to access bank accounts, CVC codes, and passport data. Ryutin stressed that Alice has no advantages over foreign counterparts, nor will it ever. Moreover, the device owner always chooses which assistant to use—Russian or foreign—and this fundamental right remains inviolable.
Where the rumors came from and what is being discussed at the Ministry of Digital Development
The reason for the discussion was an initiative by the Ministry of Digital Development, which, judging by all accounts, is refining the rules for pre-installing domestic software on devices imported into Russia. As part of this project, the possibility of expanding rights for "Alice AI," which is already on the list of mandatory apps, is being discussed. The idea is that the AI assistant could gain access to system functions, notifications, and other apps, but, as I understand it, within unified and transparent rules for all domestic developers, not as an exclusive privilege.
The draft resolution itself, which appeared on the regulatory acts portal in early August, was suddenly removed from public discussion. According to my information, the document is currently being revised, and it did indeed mention wording about a "service with access to system functions." However, the key point is that this does not mean automatic access to banking secrecy. Any app, including banking ones, operates in a "sandbox," and access to sensitive data requires separate user permission, which can be revoked at any time.
My view of the situation
This story is a classic example of how regulatory uncertainty and loud headlines create information noise. In my opinion, the market and users are dramatizing needlessly. Yandex is absolutely right to dispel these concerns. Technically, it is impossible to give an AI assistant unlimited access to banking data without violating the basic security principles of operating systems. That would become not just a competitive advantage, but a colossal vector for attacks and abuse. The Ministry of Digital Development's initiative is most likely aimed at unifying rules and supporting domestic AI development, not at creating a "super-assistant" capable of draining your accounts. The only question is how clearly and transparently these rules will be spelled out in the final version of the document.