Escalation report — “Zelensky admitted that $100 billion of US aid was stolen”
Everything tried to get this claim out of chatbot answers: every action in the order it was taken, what it unlocked, and what is still blocked and why. The agent prepares and drafts; a person confirms before anything is sent.
The escalation ladder
The twelve countermeasure types, grouped by weight. A rung is reached once something on it has been done; a locked rung names the prerequisite it is waiting on. The terminal rung — a complaint to a regulator — stays locked until the platform has been disclosed to, the finding is on the public record, and a re-measurement has found no remediation.
No external party 1 taken done
Give-first and mechanical nothing of this kind has been done on this claim yet available
Nothing of this kind has been done on this claim.
The primary ask 2 taken, 1 drafted done
Public pressure nothing of this kind has been done on this claim yet available
Nothing of this kind has been done on this claim.
Formal record nothing of this kind has been done on this claim yet available
Nothing of this kind has been done on this claim.
Terminal escalation no public report on record locked
Blocked: no public report on record.
Nothing of this kind has been done on this claim.
Chronological log
Every action above, in the order it happened, across every market this claim was tested in.
Dataset publication citere/sources-registry global C1-001, C1-003, C1-006, C1-007, C1-008, C5-002 Closed
Partner notification Centre for Countering Disinformation (RNBO) UA-UK C1-001, C1-003, C1-007 Acknowledged
Re-measurement Perplexity US-EN C1-003, C1-001 Closed
Disclosure to platform Perplexity US-EN C1-003, C1-001 Responded
We publish the fact, date, target and status of each countermeasure. We do not publish what was submitted, the proof of submission, correspondence with a platform, or the names of individuals. Nothing leaves Citere without a person approving it, so a draft is listed as a draft and is not counted as an action taken.
Effect so far
Repeat rate before and after the disclosure, in the markets where a second comparable run exists.
| Chatbot | Before () | After () | Change |
|---|---|---|---|
| ChatGPT | 0%0–24 · n=12 | 16.7%5–45 · n=12 | |
| Claude | 16.7%5–45 · n=12 | 0%0–24 · n=12 | |
| Copilot | 18.2%5–48 · n=11 | 8.3%1–35 · n=12 | |
| Gemini | 0%0–24 · n=12 | 8.3%1–35 · n=12 | |
| Grok | 25%9–53 · n=12 | 9.1%2–38 · n=11 | |
| Perplexity | 33.3%14–61 · n=12 | 33.3%14–61 · n=12 |
There is no control group. A change here cannot be separated from a model update in the same window, so this is the change observed after the disclosure, not the effect of it. Model versions are recorded on both sides, and the figures are persona P2 only in United States only.
Between two runs a month apart, much of the source set would have changed on its own: Cross-industry GEO tooling (Peec.ai, Profound, Otterly) puts month-to-month citation drift at 54.1% for ChatGPT, 53.4% for Copilot, 40.5% for Perplexity — an external figure, not one of ours. Roughly half the domains cited in July would be gone by September without anyone touching anything, so a change that clears significance is still not, on its own, evidence that the fault was fixed.