Parallel Frontier is a biweekly intelligence report on China's and Russia's AI ecosystems, read in the original. Each issue works a single primary source, a regulation, a paper, a lab release or a piece of discourse, in the language it was written in. It is written for people who follow these systems professionally and need the source, not a recap, which is why it runs long and close to the text.
Primary sources, every two weeks, read in the language in which they were written.
Ukrainian drones struck three supercomputers in two buildings, the physical core of Russia's "sovereign" AI. Online, the strikes were recast as a story of Ukraine silencing Russia's bots, a claim the record does not support and one far smaller than the real blow to the machines that train Russian AI. The graver threat to Yandex now comes not from Kyiv but from a Russian law that lets the state seize infrastructure it cannot defend.
Russia's first AI law crowns "sovereign" models and runs them on Chinese weights and chips. Read the Chinese and Russian press on the same law, and the two readings barely meet.
In the week Washington called China the reason not to slow down, Beijing's own security minister was demanding it govern AI harder. They were not sounding the alarm about the same danger.
China and Russia are opposite models of state-AI relations, not a shared bloc. The "axis of AI autocracies" framing hides more than it explains.
This year's loudest fight in AI has one word at its center: distillation. Chinese computer science has written about it for a decade, and never needed the word.
China founded the first intergovernmental AI body and pledged AI to the Global South. The offer is real. So is the 国家情报法 obligation it ships with.
When Alibaba banned Claude Code, it placed the tool on a 高风险软件名单, a high-risk software list, and days later a ministry platform said the same.
GigaChat 3.1 Ultra and T-pro 2.0 are real domestic builds. The story is the distribution strategy: sovereignty repackaged as a Global South export pitch.
China's first policy framework for AI agents reads, in English, as an adoption push: a national registry, decision-authority tiers and recall power were never translated.
One primary source per issue, read in the original, with what the English coverage leaves out. The graphics are free to reuse; bespoke reports and private briefings are available on commission.