About
Most English-language analysis of China's and Russia's AI reads a translation of a translation: a wire-service paraphrase of a government brief, a paper counted as "Chinese" because an author has a Chinese name. Parallel Frontier reads the originals. Every regulation, paper, lab release and piece of discourse in the language it was written in, and shows what the English-language version leaves out.
It does this at two scales.
A person, reading closely. A biweekly intelligence report that takes one primary source from China or Russia and reads it in the original, in full, with what the English coverage misses.
Read the issues →
A machine, reading at scale. Evidence-graded tooling that reads every new AI paper, every name in every script, every affiliation and funder, each call graded by its evidence.
See the tools →Same question from both ends: who is building China's and Russia's AI, who pays for it, what are they actually saying, and how do we know. The report answers it one source at a time, in depth; the tools answer it across the whole record, with a measured error rate. The name nods to 平行智能 ("parallel intelligence"), a real term in Chinese AI research: parallel frontiers, read together, not just one.
Parallel Frontier is written for people who follow China's and Russia's AI professionally and need the source, not a recap. That shapes everything about it. Each issue runs long, formal, and close to the primary text, because the reader is working from it, not reading it to keep up over coffee. The depth and the register are the product, not a house affectation: if the English-language summary were enough, there would be no reason for this to exist. You are meant to be able to act on it, cite it, and check it against the original yourself.
Parallel Frontier is written and built by Caroline Swartz. Chinese language & politics (NYU), Russian politics MSc (KCL), Mandarin SIGINT spanning defensive and offensive cyber operations, security operations leadership, short- and long-form reporting in five languages, AI safety research, and an MA in computational linguistics in progress (UCL).
China and Russia are usually covered separately, if Russia is covered at all, and the separation hides more than it reveals. China's AI ecosystem is industrial-scale, state-directed, and increasingly a rule-setter in its own right. Russia's is smaller, resource-constrained, and opportunistic: closer to a policy-taker watching what China and the West do, sometimes borrowing from both. Read side by side, the comparison surfaces what neither case shows alone: which governance instruments move between authoritarian systems, where they diverge, and what each borrows from the other's playbook.
The report is published in full and free, and the tools' source is private, available for a walkthrough on request. Beyond both, I take on a small number of commissions each quarter: bespoke reports, private briefings, standing advisory, and commissioned analysis of a corpus, read from the primary sources in the original languages. Enquiries: reports@parallelfrontier.ai.
Every card and chart in the report is free to use in your slides, reports, lectures, and threads. A link back is appreciated, never required. Released under CC BY 4.0.
- This publication uses a fixed lexicon: remote (<10%), unlikely (10–35%), roughly even (35–55%), probable (55–75%), likely (75–90%), almost certain (>90%).
- Confidence (low / moderate / high) describes the strength of the evidence base, not the probability.
- Errors are corrected in the next issue and in the archive, marked.
- Real-time news, grand-strategy punditry, or translation-as-a-service. The explanation layer, every two weeks, and the measurement layer, continuously.