AI Ethics Rules for Immigration Attorneys | Rafael Research
Book a call
← Research<br>Applied AI Research
AI in Immigration Practice: What Attorneys Are Actually Required to Do
What the ABA, EOIR, and a dozen state bars actually require of<br>immigration attorneys using AI — read from the primary documents,<br>not secondhand summaries, with the reliability research to back it up.
Rafael Research · August 2026
AI adoption is accelerating across the legal profession, including<br>immigration practice, but governance has not kept pace. Informal use<br>is widespread, but formal, written AI-use policies remain the<br>exception rather than the rule at most firms — which means the<br>overwhelming majority of the profession is using these tools with no<br>documented answer to the basic questions: what can go in, what needs<br>a client's sign-off, who's liable when the output is wrong.
Those questions now have real answers. The ABA, several state bars,<br>and EOIR have now begun translating those existing duties into<br>AI-specific guidance. This piece pulls together what the actual<br>rules say, what the reliability research actually found, and what<br>that adds up to in practice — reading the primary documents<br>directly rather than working from summaries, since on a subject<br>this consequential the gap between "someone paraphrased it" and<br>"here's the sentence" tends to matter.
What the ethics rules actually say
The ABA's first opinion on generative AI, Formal Opinion 512, sets<br>the national baseline by tying five existing duties to specific AI<br>conduct.1 On competence, it doesn't<br>ask lawyers to become engineers — just to have "a reasonable<br>understanding of the capabilities and limitations of the specific<br>GAI technology" — but it's blunt about what happens without that<br>understanding: "uncritical reliance on content created by a GAI<br>tool can result in inaccurate legal advice," and the lawyer remains<br>fully responsible regardless. On confidentiality, the opinion<br>requires informed client consent before inputting case information<br>into a self-learning tool, and it specifically closes a loophole a<br>lot of firms are currently relying on: "merely adding general,<br>boiler-plate provisions to engagement letters purporting to<br>authorize the lawyer to use GAI is not sufficient." If your<br>engagement letter has a generic AI clause and nothing more, it<br>likely doesn't clear the ABA's own bar. On candor, the opinion is<br>equally direct — AI output "must be carefully reviewed to ensure<br>that the assertions made to the court are not false," and even an<br>unintentional misstatement can trigger a misrepresentation finding<br>under Rule 8.4(c). On fees, it gives a concrete example: fifteen<br>minutes spent prompting a tool, plus the time spent reviewing the<br>draft, is billable; time "saved" is not, and neither is the time a<br>lawyer spends learning to use the tool in the first place.
EOIR wrote its own memo for immigration proceedings specifically,<br>and it's thinner than most practitioners assume.2<br>There's no blanket ban on attorneys using AI, and — more<br>surprising — no mandatory national requirement to disclose that you<br>used it. What EOIR did keep is a disciplinary hook: submitting<br>hallucinated or fabricated AI content exposes an attorney to<br>sanctions under 8 C.F.R. § 1003.102(c) for "knowingly or with<br>reckless disregard offering false evidence," with the memo pointing<br>to ABA 512 as persuasive authority on the underlying standard.<br>Individual immigration judges remain free to impose their own<br>disclosure requirements through standing orders, so "EOIR doesn't<br>require it nationally" isn't the same as "no judge in front of me<br>will ask." The memo is explicitly a placeholder — EOIR said it<br>expected the Department of Justice to issue department-wide<br>generative AI guidance by roughly the end of 2025, and as of this<br>writing there's no public confirmation that guidance has actually<br>been issued, which is worth checking before treating PM 25-40 as<br>the final word.
State bars have moved faster and further than the ABA in a few<br>notable ways. Florida's opinion contains the strictest billing<br>language of any jurisdiction reviewed, barring lawyers from billing<br>in any way that would "falsely inflate the lawyer's billable hours"<br>and requiring AI output to be reviewed the same way a lawyer would<br>review a paralegal's work.3 Texas's<br>opinion — finalized in February 2025 as Opinion No. 705, after<br>circulating as a draft the previous year — flags a specific and<br>easy-to-miss risk: "many generative AI tools invite a 'conversation'<br>in which the lawyer... will explain relevant facts," which can leak<br>more confidential detail than a lawyer intends if the tool retains<br>or trains on what's typed in.4 New<br>York's task force coined a term worth knowing —<br>"techno-solutionism," the assumption that AI automatically solves a<br>profession's access-to-justice problems — and used it as a caution<br>rather than a dismissal, going so far as to suggest that refusing<br>to adopt genuinely useful, well-vetted tools could itself become...