By any historical measure, moments like this do not announce themselves loudly. They arrive wrapped in professional letterhead, careful language, and procedural restraint. They look reasonable. They sound responsible. And by the time their consequences become clear, the damage is already embedded in the system.
That is why the National Court Reporters Association’s recent submission to the Office of Science and Technology Policy—titled Regulatory Reforms for AI Innovation in Federal Court Proceedings—has left many court reporters unsettled, disappointed, and deeply uneasy.
On its face, the document does not endorse artificial intelligence as a replacement for stenographic court reporters. It affirms stenography as the “gold standard.” It acknowledges AI’s accuracy failures, bias risks, and lack of accountability. It urges caution. It calls for study.
And yet, for all its professionalism, the submission reveals a troubling truth: in a moment requiring firm opposition, it chooses procedural accommodation.
This is the good, the bad, and the ugly of that choice—and why it matters more than many may yet realize.
The Good: What the Submission Gets Right
The filing is not careless. It does not blindly cheerlead automation. It accurately identifies real technical and ethical problems with AI-generated transcripts, including homophones, overlapping speech, numerical inaccuracies, and bias across demographic groups. These are not theoretical concerns; they are documented failures already appearing in courtrooms.
The submission also rightly emphasizes accountability. Human court reporters attach certification pages. They can be subpoenaed. They can testify to the integrity of the record. Artificial intelligence cannot do any of those things. That distinction is not cosmetic. It is foundational to how courts authenticate evidence.
Most importantly, the document states—on the federal record—that stenographic court reporting remains the gold standard. That phrase matters. It can be cited. It establishes institutional position. It prevents misrepresentation by vendors claiming professional endorsement.
These are not trivial contributions. They are necessary truths.
But necessity is not sufficiency.
The Bad: What the Submission Fails to Do
The document’s central weakness is not factual. It is strategic.
From the opening paragraph forward, the submission assumes that AI integration into court proceedings is inevitable. The only question posed is how quickly and under what guardrails it should occur. This framing quietly surrenders the most important argument court reporters have: that some functions of the justice system should not be automated at all.
Instead of drawing a bright line, the submission asks for a task force. An 18–24 month study. A measured approach. Consultation before implementation.
That may sound prudent. In practice, it functions as a stall while deployment continues.
Courts are not waiting. Vendors are not waiting. Digital recording systems are already normalized in many jurisdictions. “Pilot programs” quietly become permanent fixtures. And once an inferior system is embedded, courts rarely reverse course.
The filing also stops short of demanding what would actually protect the profession—and the public: mandatory use of licensed stenographic court reporters for the creation of the official record. Without that demand, the “gold standard” becomes merely aspirational.
A standard that is optional is not a standard. It is a suggestion.
The Ugly: What This Signals About the Future
The most troubling aspect of the submission is what it signals implicitly.
By positioning court reporters as stakeholders in AI deployment rather than as guardians of the official record, the profession is subtly recast as legacy infrastructure—valuable, but negotiable. Advisory, not essential.
This is the language of managed decline.
History is full of professions that made this mistake. Newspaper journalists were told automation would “free them” to do higher-level work. Travel agents were told online booking tools would expand their reach. Each was invited to help manage the transition. None retained control once the transition was complete.
The justice system is not a market. The court record is not a convenience. And yet this submission treats AI integration as an innovation challenge rather than a constitutional one.
There is no discussion of appellate risk. No reference to due process. No confrontation with the reality that an uncertified, unauditable record undermines the fairness of proceedings long after the hearing ends.
The ugliness lies not in what is said, but in what is normalized by omission.
A Line-by-Line Rewrite: How This Could Have Been Stronger
Below are representative examples of how key sections could have been rewritten—not rhetorically, but substantively.
Original:
“We respectfully urge the Administration to establish a Department of Justice Task Force on AI in the Justice System before implementing any regulatory changes…”
Stronger:
“We respectfully urge the Administration to impose an immediate moratorium on the use of AI-generated transcripts as the official court record until Congress and the judiciary have determined whether such records satisfy constitutional due process, evidentiary reliability, and appellate review requirements.”
Original:
“Federal policy must prioritize stenographic court reporting as the gold standard…”
Stronger:
“Federal policy must mandate licensed stenographic court reporters as the exclusive creators of the official court record in federal proceedings, with any digital or AI tools limited strictly to non-authoritative, supplemental functions.”
Original:
“AI struggles with homophones, number figures, overlapping speech…”
Stronger:
“These failures are not mere technical defects; they constitute material alterations of testimony that expose litigants to reversible error, sanctions, and loss of appellate rights.”
Original:
“AI transcripts do not have certification pages…”
Stronger:
“An uncertified transcript cannot satisfy evidentiary standards for authenticity, chain of custody, or judicial notice. Any system that produces such records is incompatible with the justice system’s obligations.”
A Member Resolution Calling for a Harder Line
Resolution of the Membership of the National Court Reporters Association
WHEREAS, the official court record is a foundational safeguard of due process and appellate review; and
WHEREAS, licensed stenographic court reporters are uniquely trained, certified, and legally accountable for the accuracy and integrity of that record; and
WHEREAS, artificial intelligence and automated speech recognition systems lack certification, accountability, and the ability to testify to authenticity; and
WHEREAS, incremental or “pilot” deployment of AI systems risks permanent erosion of record integrity;
BE IT RESOLVED, that the National Court Reporters Association shall oppose the use of AI-generated transcripts as the official record in any judicial proceeding;
BE IT FURTHER RESOLVED, that NCRA shall advocate for mandatory use of licensed stenographic court reporters in federal and state courts;
BE IT FURTHER RESOLVED, that NCRA shall reject “capture-method neutrality” where such neutrality compromises admissibility, accountability, or constitutional rights;
BE IT FURTHER RESOLVED, that NCRA shall communicate these positions clearly and publicly to courts, lawmakers, and the legal community.
What a Stronger Alternative Submission Should Have Said
A stronger submission would not ask for permission to study harm already documented. It would state plainly:
- AI-generated transcripts are inadmissible as official records
- Uncertified records undermine due process
- Courts adopting such systems assume legal risk
- Licensed reporters are not optional infrastructure
It would frame stenography not as tradition, but as constitutional compliance.
The Bottom Line
This filing is not malicious. It is cautious. But in moments of structural change, caution can become complicity.
Court reporters do not merely document justice. They make justice reviewable. And a profession that forgets that role—however politely—risks being written out of the future it helped build.
If the record is to remain trustworthy, advocacy must be firmer than this. The moment demands clarity, not accommodation.
Disclaimer:
This article reflects the author’s analysis and opinion regarding policy positions and public statements related to court reporting, artificial intelligence, and the integrity of the judicial record. It is based on publicly available information and professional experience within the legal system. Nothing herein is intended as legal advice, nor as an assertion of fact regarding the motives, intent, or conduct of any individual or organization. Readers are encouraged to consult primary sources and applicable law when evaluating these issues.
STENOIMPERIUM 