“YaLM 100B is a GPT-like neural network for generating and processing text. It can be used freely by developers and researchers from all over the world.” This is the introduction to a ground-breaking open source technology developed by Yandex on and made public on its platform “GitHub” (https://github.com/yandex/YaLM-100B) on 23 May 2022.
The technology is available in the English and Russian languages, leverages 100 billion parameters and is reportedly the largest GPT-like model for Russian language. It can write different types of texts based on self-learning from sources in Russian (75%) and English (25%). It is published under an Apache 2.0 license comprised of research and commercial use rights.
The Apache license was developed and approved by the Apache Software Foundation, a US charitable organization with the mission to provide software for the public good, which covers terms and conditions for use, reproduction, and distribution of works. Licenses granted under Apache 2.0 are perpetual, worldwide, non-exclusive, not-for-charge, royalty-free, irrevocable and allowing reproduction, public display, public performance, distribution, and sublicense of works. Notwithstanding Apache being born in the US, it is used world-wide by developers of all characterizations and claims to be the world’s largest open source institution.
Yandex is a commercial, publicly traded company founded by Russian developers in 1997 and is one of the major IT players mainly operating in Russia. In a complex global society, where technological progress, commercial interests and geopolitical drivers are increasingly had to reconcile, Yandex’s choice to make freely available to the general public an extremely advanced product may sound surprising and deserves reflection in the broadest sense. This is also a good opportunity for a high level analysis of how a general open license fits within the legal framework of computer licenses laid down by the Russian Civil Code.
Articles 1235 and 1286 provide general rules in the matter of licenses for copyright protected works. Special provisions related to software licensing are rather limited.
Parts 4 and 5 of article 1286 allow the rightholder to grant under a license agreement rights for use, besides those mentioned in article 1280, as a program user rights. Such agreement may be concluded in simple form and is as a rule an “adhesion contract” drawn up by the rightholder.
The legal concept of open license was added by article 1286.1 of the Civil Code in 2014, which contemplates that a work of science, literature or art can be subject of an open license. Computer programs are literature works. An open license is an adhesion contract, which can be concluded by the licensee in a simple form even by performing certain actions. All terms and conditions of the open license agreement shall be made available to the general public and made clear to the licensee before the first use of the work.
The licensor may grant the right to create new intellectual property on the basis of licensed work (derivative work). This means that the licensor makes a public offer to enter into an existing open license to all members of the public, who may in the future use new IP rights created on the basis of (i.e. by derivation from) the licensor’s work on the same terms and conditions as the open license.
The general rule also provides that unless otherwise provided by the open license agreement, the right is deemed granted free of charge and is valid on a worldwide basis.
The law moreover provides for different default license terms to software and other copyright works, setting 5 a year term for all works vs. author’s lifetime term for computer program rights.
Besides, the law reserves the right of the licensor to withdraw the open license, if the licensee grants rights to the licensed work or its new derivative work on terms and conditions other than those provided by the open license.
To sum up, although the governance of software open source licensing under Russian law is not extensive, it yet fits within the established systematic approach to copyright licenses.
IT is by definition global and, according to general information, Russian developers widely use the GitHub platform both to share and obtain open source license software. From the end of February 2022, developers of open source software started restricting access to their works via GitHub to developers and users from Russia. Later, in April 2022, in compliance with the US technological sanctions against Russia, GitHub banned dozens of accounts from Russia, including both developers and major Russian banks. After that, Russian officials sources announced that a Russian repository of open source software shall be launched before the end of 2022 and shall be available for use by Russian and foreign developers alike.
Meanwhile, Russian developers built and expanded ties with Chinese counterparts, who are eager to attract them and their open source software to the Chinese open source platform OSCHINA and Chinese repository Gitee.
Another recent measure by the Russian authorities has been to provide open licenses to all software codes created under Russian State funding and owned by the Russian Federation. This should result in the grant of open licenses to a potentially indefinite number of users, based on three main drivers:
- software to be placed in a cloud repository system and made available for use by all project members,
- creation of a Russian storage site for keeping track of all versions of software (similar to GitHub or Gitee), which shall be open to all members irrespective of location,
- collection and analysis of knowledge and recommendations on the fundamentals of regulation of open source licensing and the use of State owned software.
So, Yandex’s example may well have marked a new wave notwithstanding the adverse geopolitical conditions, and innovative legislation on open source software licensing is likely to be launched in Russia in the near future.