The promulgation deficit
The condition of a profession in which the outcome standards for AI-assisted work are old and settled (do not file what does not exist), the method standards (what supervision, verification and record constitute adequate professional control of a generative system) have been promulgated by no governing body, and the missing method standard is therefore being written retrospectively, one sanction order at a time, by the tribunals that punish its breach.
When this decides something for you
A Bar Council, a medical council or an engineering board is asked what its members may do with generative AI. It issues guidance: your existing duties of competence and integrity apply. A practitioner reads it and still does not know whether machine cross-checking satisfies the duty of verification, whether delegating to a junior who used a chatbot breaks the chain of supervision, or what record of checking a tribunal will later accept as diligence. The practitioner discovers the answers when sanctioned. This page names that condition, because a norm discoverable only by being punished under it fails the most basic rule-of-law test a profession can set itself.
The question
Why are the binding norms on AI in the professions being authored by disciplinary cases rather than by the bodies that hold the power to make rules?
The premise
Every mature profession runs on two kinds of standard. Outcome standards state what finished work must not contain: a pleading must not cite authorities that do not exist; an audit must not certify accounts that misstate. Method standards state how work must be done so that those outcomes are reliable: what checking, what supervision, what independence of judgment constitutes adequate control. Outcome standards are ancient. Method standards are written when a new method arrives.
How it works
Generative AI did not change the outcome standard. No advocate in any legal system was ever free to invent a precedent. What it changed is the economics of breaching the standard: fabrication that once required deliberate fraud now requires only ordinary negligence at scale, because the machine produces fluent, formatted, confident falsehood as a by-product of normal operation.
For that new method, the method standard is promulgated nowhere. No bar council, medical council or engineering board has enacted a binding rule stating what supervision of a generative system constitutes adequate professional control: what must be independently verified, against which class of source, by whom, with what record. What exists instead is enforcement. Courts and regulators, confronted with fabricated authorities in filed documents, sanction practitioners one order at a time, and each order writes a fragment of the missing standard after the event: this reliance was inadequate; that verification would have sufficed. Enforcement runs years ahead of promulgation. The gap is the promulgation deficit.
Two objections meet the diagnosis and both fail. First, that the duty always existed, so nothing needs promulgating. The outcome duty existed; the method standard did not, because the method it must govern did not exist, and "do not file fabricated authority" says nothing about whether machine cross-checking discharges the duty. Second, that guidance is everywhere. It is, and almost all of it is non-binding, fragmented and outcome-focused. A guidance note that says "your existing duties apply" restates the outcome standard in a new context; it does not enact the method standard whose absence is producing the sanctions. Mapping is not promulgation.
Nor is the method standard mysterious. One version of it has been specified and applied in a live High Court practice: a staged method reserving framing, verification and strategic judgment as human-only acts, paired with a four-step test of responsible use. The point is not that this instrument is the answer. It is that instruments of this kind are writable now, replicable now and promulgable now by any body with the authority to bind, which makes the deficit a choice rather than a fate.
One case
Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd, 2026 INSC 668, Supreme Court of India, 2 July 2026. Having set aside tribunal decisions that rested on precedents "obtained through its own research" rather than cited by counsel, the Court held that "mere declaration of prohibitory action is not sufficient; there must be a consequential action following accountability", and directed the Bar Council of India, "being the apex statutory body, to constitute a committee and deliberate... and prescribe a guiding principle to prevent such occurrences, along with the disciplinary action that will follow a violation of the norms". That is a court naming the deficit and ordering the body with rule-making power to close it: promulgate the method standard, and attach discipline to it, so that enforcement stops writing the rule.
The rule
Where the outcome standard is old and the method standard is unwritten, sanction orders will write it. The body that holds the power to promulgate either does so or cedes its rule-making to the tribunals that punish its members.
Not to be confused with
The deficit is not an absence of law. The outcome standard binds, and practitioners are rightly sanctioned under it. Nor is it an absence of guidance; guidance is abundant. It is the absence of a binding statement of method from the body whose function is to state it, and it is a condition of every profession in the AI transition, not of lawyers alone.
Provenance and method.
Where this was published
- Shivam Shukla
- The Power to Say No: Activating the Missing Accountability Layer of the Solidarity Stack
- SSRN 7454698 (2026)
- section 2
Changelog
- 2026-09-25, version 1. First published.