ChatGPT Summaries, Uncertified Translations and an AI Researcher’s EB-1A Appeal
An application’s evidence must establish the underlying facts, even when its summaries sound authoritative. In In Re: 43945732, a graduate student researching causal artificial intelligence relied on awards, memberships, patents, publications, project work, income, and conference posters. The AAO identified problems with evidentiary fit and documentation, including unverified ChatGPT-sourced background material and uncertified machine translations. It dismissed the appeal because the record did not establish at least three initial EB-1A criteria.
The August 25, 2026 nonprecedent decision contains twelve pages of analysis and footnotes. Its reasoning concerns this record’s reliability and proof, not a categorical prohibition on using AI during case preparation or a rule excluding graduate students from extraordinary-ability classification.
Award and membership claims required evidence of the specific requirements
For awards, the AAO distinguished the reputation of an organizer or conference from recognition of the particular prize. Some materials described competitions without establishing the recognition or receipt of the claimed award. Footnote 2 also identified an unexplained inconsistency concerning two prizes for a submission. See pages 3–4.
Footnotes 3 and 4 noted ChatGPT.com source URLs on background documents and translations. The AAO found that reliability had not been demonstrated. For memberships, graduate-student status or senior-membership conditions did not establish outstanding achievements judged by recognized experts; footnote 5 identified the same sourcing issue in membership background material. See pages 4–5.
These were proof problems. A generated description of an association’s admission requirements does not replace the authoritative requirements and evidence showing how the relevant membership tier applies.
Translation certification was a separate defect
8 C.F.R. § 103.2(b)(3) requires a full English translation certified as complete and accurate, together with certification of the translator’s competence. The decision quotes the requirement that the translator certify “that he or she is competent to translate from the foreign language into English.” See page 6, footnote 6.
Patent translations bore Google-translation language without proper certification. Salary-related documents and translations were also blurry or illegible, with a separate certification problem described in footnote 8. Machine assistance did not relieve the petitioner of the document-quality and certification requirements. The decision did not hold that a properly certified translation must be rejected merely because software assisted its preparation.
Patents and citations did not establish major significance
The record did not show that the U.S. patent application had been granted or attracted significant attention. Chinese patent records and an adviser’s letter did not sufficiently establish field-level significance. The AAO also found the explanation of the petitioner’s own contributions inadequate for heavily coauthored work. Citation counts and brief references did not, without context, establish major significance. See pages 6–8.
The analysis distinguished originality, individual contribution, and significance. A patent’s existence, an aggregate citation count, or a prediction of future usefulness did not automatically establish all three.
Project participation, pay comparisons and posters left other gaps
A project report named another individual as leader and listed the petitioner among graduate students. The evidence did not sufficiently establish his claimed leading or critical organizational role. Government funding of a project did not itself prove that role. See pages 8–9.
For high remuneration, the petitioner characterized his income as research-assistant compensation but compared it with scholarships and grants. The AAO found the comparison inappropriate for the claimed work and pay. Conference certificates and photographs beside posters also did not sufficiently establish the proposed comparable-evidence claim for artistic displays. See pages 9–11.
What the AAO expressly left undecided
The AAO reserved judging and scholarly authorship because even favorable findings on both could not yield three criteria after the other claims failed. It also reserved final merits. Accordingly, the decision should not be summarized as an AAO finding that he had never judged work or authored scholarly articles, or as an independent final-merits denial. The appeal was dismissed. See pages 11–12.
The practical preparation task is to trace each factual claim to reliable, legible evidence, supply required certifications, identify the petitioner’s contribution, and explain the correct comparison. AI output can assist organization, but it cannot establish facts by assertion. Special circumstances, the complete record, and applicable later law require individual review.
Legal Disclaimer. Originally published: 11/01/2026. This article provides general information only and is not legal advice for any particular matter. Laws, policies, procedures, fees and agency practices may change; their application depends on the facts and jurisdiction. Consult a qualified attorney about your circumstances before acting on this information. Reading this article or contacting Blumsack & Canzano does not, by itself, create an attorney-client relationship. Past results do not guarantee future outcomes. This material may be considered attorney advertising in some jurisdictions.
