Which Chinese AI researchers face restriction in the next twelve months?

The signal is in the specification. The surprise is a reading failure.

Chinese AI researcher restrictions ran through three waves between 2020 and 2026. The first wave — Entity List actions in 2020 and 2021 — targeted Huawei, supercomputing centers, and quantum-AI institutes, with researcher-level effects concentrated in joint US–PRC programs. The second wave landed October 7, 2022: a US-persons restriction that drove low-hundreds of US-passport engineers out of YMTC, CXMT, SMIC, and the chip-design houses around them. The third wave — the one most recently legible — runs in the other direction: passport custody and pre-travel review extending from state labs into DeepSeek, Moonshot, Zhipu, Alibaba DAMO and Tongyi, with the May 2026 Bloomberg dispatch and a same-week Tech Times confirmation as the founding press citations.¹

The historical work exists in slices. MacroPolo's AI Talent Tracker mapped stocks and flows from 2020 through its 2024 discontinuation. CSET tracks aggregate inflows and outflows. Stratfor and Recorded Future publish episodic risk briefings on named individuals from behind subscription walls. None of these is a dated, public, per-researcher forward score. None is Brier-scored.²

Forward Indicators ships the per-researcher score: prior public, features versioned, three baselines run on the same scoreboard with Brier reported, and the editorial guardrails in §VIII binding on every published number. v0 is five profiles. v0.1 ships at twenty-five. The point of v0 is not the prediction — the point of v0 is that the prediction now exists in a form that can be wrong in public.

Five moving parts. Each is versioned. Each ships.

The pipeline implements the Forward Indicators spine on a single corpus: a public-source profile for each researcher, labelled against the six-year restriction record, scored by a logistic baseline, and reviewed against two trivial baselines on every release.

Six hand-engineered features — institutional tier, research area, US co-authorship density, citizenship-and-visa status, CCP-affiliation signals, and recent mobility — all derived from public lab pages, arXiv, conference proceedings, or announced funding rosters. Each feature carries an empirical band drawn from a specific §1 case. Two further features (stated geopolitical alignment, mentor lineage) are scoped for the v0.2.0 LLM-extractor lane and are not in the v0 scorer.

Seventeen primary restriction events from RESEARCH.md §1 plus a long tail of inductive instances expanded from the same §1 patterns into an event-cluster corpus. Positive instances are synthesized populations affected by each event (e.g., “YMTC US-passport engineers”); negative instances are peer researchers at the same labs in the same time windows who do not appear in restriction reporting. Total labeled n = 1,022 with a 22% positive rate.

A logistic regression on the standardized hand-engineered features is the v0 scorer. Coefficients are stable in sign across reseeds: mobility loads heaviest (+1.68), area and tier moderately positive (+0.68, +0.48), US density mildly protective at this point in the regime cycle (−0.18), CCP-signal count mildly negative (−0.26) once tier and area are controlled for. Calibration is approximate; the headline metric is Brier, not point accuracy.

Three baselines run alongside on every release: a constant predictor at p = 0.20, the last-observed positive rate from training, and the production logistic regression. The logistic lifts Brier by roughly half versus the trivial baselines on the held-out split (0.088 vs ~0.171), and AUC from 0.50 to 0.91. Both metrics are stamped to backtest_v0.json on every pipeline run.

Brier score is the headline. Alongside: AUC-ROC, calibration curve (reliability diagram), precision @ top-20 and top-100, and lift over each baseline. The reliability diagram — not point accuracy — is the chart on the scoreboard.

What we will and will not publish.

Public profiles only. Every input field is recoverable from a public lab page, arXiv, conference proceedings, or an announced funding roster. No private personal information enters the pipeline.

Pseudonyms by default in v0.Real names enter the registry only after (a) a documented restriction event is available as a citable anchor for the named individual, or (b) an external advisor review explicitly approves a named profile on the basis of the researcher's own publicly-stated positions. v0 ships with zero real-name case studies.

No published score above 0.80 without a supporting evidence pack. The v0 scorer clamps at 0.80 by construction; this guardrail will remain in force through v0 and v0.5.

Right to correction. Researchers may request data corrections at any time. Corrected records carry a versioned diff; the prior version remains visible in the append-only log. To request a correction, write to corrections@forward-indicators.org.

Single-source claims marked. Every fact that rests on a single press source is marked on the row. Every fact for which a specific URL or document identifier is not yet pinned down is flagged [unverified, source pending] and resolved on the punch list before the next release.