# AI and the Duty of Technology Competence

> Model Rule 1.1 and Comment 8 now reach AI. A concrete, verifiable checklist for what "reasonable understanding" of any AI tool actually requires of a lawyer.

[Home](https://rankshieldlegal.com/) / [Blog](https://rankshieldlegal.com/blog/) / Legal AI AI ethics & competence
# The New Duty of Technology Competence: What Model Rule 1.1 Now Expects You to Verify About AI
The duty of competence has always covered more than doctrine. Under Model Rule 1.1 and its technology comment, and as elaborated for AI by ABA Formal Opinion 512, a lawyer who uses an AI tool is expected to have a reasonable understanding of what that tool does and does not do. This article turns that abstract duty into a concrete list of things you can actually verify, and it is informational, not legal advice.

By [Jamie Kloncz](https://rankshieldlegal.com/about/), Founder, RankShield ** 21 min read ** Published July 1, 2026

The short answer is that Model Rule 1.1 now expects you to understand, at a reasonable level, the AI tools you use on client matters, and to check their output rather than trust it. The duty of competence has always required a lawyer to keep up with the tools of the trade. Comment 8 to Rule 1.1, often called the technology competence comment, made that explicit for technology generally, and ABA Formal Opinion 512 elaborated it for generative AI in particular. You do not have to become an AI expert. You do have to have a reasonable understanding of the capabilities and limitations of the specific tools you use, including where client information goes and how outputs are checked.
This piece is written by a founder who builds verification tools for law firms, not by an attorney, and it is general information rather than legal advice on any specific matter. The aim is to take a phrase that sounds abstract, the duty of technology competence, and turn it into something concrete: a short list of things you can actually verify about any AI tool before you rely on it. Where the guidance is fact-specific or varies by jurisdiction, this article says so, and it points you back to your own counsel and your state's adopted rules for the decision itself.

## Rule 1.1 already reached technology before anyone said "AI"
Comment 8 to Rule 1.1 says a lawyer should keep abreast of the benefits and risks of relevant technology. It was written broadly, before generative AI, and most states have adopted a version of it. That is why the competence duty already reaches the AI tools a lawyer uses.
Model Rule 1.1 states the baseline duty in one sentence: a lawyer shall provide competent representation to a client, which requires the legal knowledge, skill, thoroughness, and preparation reasonably necessary for the representation [[1]](#ref-1). That has never been limited to knowing the law. Competent representation has always included handling the tools a matter runs on with reasonable care, from the filing system to the research database to the software that touches client files.
Comment 8 to Rule 1.1 made the technology piece explicit. To maintain the requisite knowledge and skill, the comment says, a lawyer should keep abreast of the benefits and risks associated with relevant technology [[2]](#ref-2). That language predates the current wave of generative AI, and it was written broadly on purpose. It does not name any particular product. It ties the competence duty to whatever technology is relevant to the representation at the time, which is why it reaches AI without needing to be rewritten.
A large majority of states have adopted a version of Comment 8, which is why the technology competence duty is not just an ABA idea but a live obligation in most jurisdictions. This article does not assert an exact count or specific adoption dates, because those vary and shift. The safe reading is that if you practice in the United States, some version of a technology competence expectation very likely applies to you, and the ABA comment is the reference point most state versions are built on. Check your own jurisdiction's adopted rule for the wording that governs your practice.

## What Formal Opinion 512 added: reasonable understanding, not expertise
ABA Formal Opinion 512, issued in 2024, took the general technology competence duty and applied it directly to generative AI. Its most useful move for a practicing lawyer is where it sets the bar. The opinion makes clear that a lawyer need not become an AI expert to use these tools competently, but must have a reasonable understanding of the capabilities and limitations of the specific AI tools the lawyer uses [[3]](#ref-3). That framing matters because it defines the duty in terms a firm can actually meet.
The word "specific" is doing real work there. The duty does not attach to AI as an abstract category. It attaches to the particular tools you deploy on client matters, each of which behaves differently. A general chatbot, a research platform with retrieval, and an enterprise system with a no-training agreement are three different risk profiles, and reasonable understanding means understanding the one in front of you well enough to judge how it should be used and where it should not.
Opinion 512 also ties competence to the duties around it, including communication under Rule 1.4, confidentiality under Rule 1.6, and supervision under Rules 5.1 and 5.3 [[3]](#ref-3). Competence is the hinge because the others depend on it. A lawyer cannot supervise the use of a tool the lawyer does not understand, cannot explain a risk to a client the lawyer has not grasped, and cannot protect confidentiality without knowing where inputs go. That is why turning competence into a concrete, checkable standard is the practical starting point.
Opinion 512 sets the bar at reasonable understanding of the specific tools you use, not at AI expertise. That is a standard a firm can meet with a defined process rather than by hiring a data scientist.

Source: ABA Model Rule 1.1 and Comment 8; ABA Formal Op. 512 Download SVG

## Reasonable understanding, drawn out into what it does and does not require
Because the standard is "reasonable understanding," the practical question is where that line sits. It is easy to read the duty as either trivially low, meaning you clicked around once, or impossibly high, meaning you understand the model's internals. Neither is right. The table below separates what reasonable understanding tends to require from what it does not, drawn from how Opinion 512 frames the competence duty [[3]](#ref-3). It is a working reference to structure the conversation with your own counsel, not a scoring system that produces an answer by itself.
Read the two columns together. The left column is about the tool's behavior at the level that affects client work: what it produces, where inputs go, and how output is checked. The right column is the deep technical machinery that the opinion does not require a lawyer to master. The duty lives in the left column, and most of it is answerable without any data science background, often by reading the vendor's documentation and contract with a critical eye.
Reasonable understanding tends to require It does not require
Knowing what the tool does and does not do, including that a general model can produce fabricated citations [3] Understanding the model's architecture, weights, or training mathematics
Knowing where your inputs go, and whether they are retained or used for training Being able to build or fine-tune a model yourself
Knowing how the tool's outputs are checked before they are relied on Reviewing the tool's source code line by line
Knowing whether the tool carries confidentiality safeguards appropriate to the data Auditing the vendor's data center or hardware
Knowing enough to supervise staff use and explain the risk to a client [3] Achieving expert-level fluency in machine learning

The duty sits in the left column. Most of it is answerable from a vendor's documentation and contract read critically, not from technical training. Guidance here is general and fact-specific; apply it with your own counsel.

## The three things worth verifying about any AI tool
Three questions cover most of it: what the tool does and does not do, where your inputs go, and how outputs are checked. The first is documented, the second is contractual, and the third is your own process. None requires being an AI expert.
Opinion 512's framing collapses into three questions that apply to essentially any AI tool a lawyer might use. They are worth stating on their own because they turn the abstract duty into something you can actually check before a tool touches client work. None of the three requires technical expertise. Each requires a specific, honest answer rather than a marketing summary.
The first is what the tool does and does not do. The clearest example, and the one that has produced sanctions, is that a general large language model can produce fabricated citations, which is exactly why cited authorities have to be verified rather than trusted [[3]](#ref-3). The second is where your inputs go: whether they leave your control, whether they are retained, and whether they are used to train the model. The third is how outputs are checked, which is your own process for catching errors before they reach a filing or a client.
These three are not equally hard to answer, and that is part of the point. What the tool does is usually documented, if you read past the sales language. Where inputs go is a contract and configuration question, answerable in writing. How outputs are checked is entirely within the firm's control, because it is your workflow, not the vendor's. A firm that can answer all three in plain terms for a given tool is a long way toward the reasonable understanding the duty asks for.

## A competence checklist you can run on any AI tool
The value of a checklist is that it makes the competence duty routine rather than something reconstructed under pressure after a problem. The steps below map to the three verification points and to how Opinion 512 frames the duty [[3]](#ref-3). Run them before a tool is used on client matters, and revisit them when the tool changes or a new one is introduced. This is a structuring aid, not legal advice, and it does not replace your jurisdiction's adopted rules or your own counsel's judgment.
Work the list in order and stop to change course at the first point where the honest answer is unsatisfactory. The goal is not to check boxes toward a predetermined yes. It is to make sure that when you rely on an AI tool, you can explain, in plain terms, what it does, what it does with client information, and how you catch its mistakes.

- **Identify what the tool actually does** Write down, in one or two sentences, what the tool produces and what it does not do, including that a general model can generate fabricated citations that must be verified [[3]](#ref-3).
- **Trace where inputs go** Determine whether inputs leave your control, whether they are retained, and whether they are used for training, from the contract and configuration rather than the marketing page.
- **Match the tool to the sensitivity of the data** Decide whether the tool's confidentiality safeguards fit the information you intend to put into it, and keep confidential material out of tools that do not fit.
- **Define the output-checking step** Set a concrete process for verifying outputs, including checking every cited authority against the actual source before anything is filed or sent.
- **Set the supervision expectation** Decide how staff use of the tool is overseen so it sits inside firm policy rather than being ad hoc, consistent with Rules 5.1 and 5.3 [[3]](#ref-3).
- **Keep a record you can show later** Note which tool was reviewed, what it does with data, and that the checking step exists, so the firm can demonstrate its handling rather than merely assert it.
This checklist is a competence aid, not legal advice. Apply it with your own counsel and your jurisdiction's adopted rules on the specific tools and matters in front of you.

## Fabricated citations are the clearest competence failure
Of everything the competence duty covers, the fabricated citation is the sharpest illustration, because it is where a lack of reasonable understanding becomes visible in a filing. A general large language model produces text that is statistically plausible, and plausible text can include case names, citations, and quotations that do not exist. This is not a tool being broken. It is a known characteristic of how general models work, and Opinion 512 treats understanding characteristics like this as part of the competence duty [[3]](#ref-3).
The reason this matters for competence specifically is that the failure mode is predictable. A lawyer with a reasonable understanding of a general model knows in advance that its citations cannot be trusted on their face, and therefore verifies each one against the actual source before filing. A lawyer who treats the model as if it were a search engine returning real results has not met the understanding the duty asks for, and the fabricated citation is the evidence of that gap. No tool should ever be described as hallucination-free, and a firm that hears that claim should treat it as a reason for more scrutiny, not less.
The practical discipline is simple to state and non-negotiable to follow: cited authorities must be verified, every time, regardless of how confident the output looks. Confidence in the phrasing is not evidence that the citation is real. For the difference between a citation that is wholly invented and one that exists but does not support the point, see [our note on fabricated versus misgrounded citations](https://rankshieldlegal.com/fabricated-vs-misgrounded-legal-ai-citations/), and for the mechanics of checking, our [guidance on verifying AI case citations before filing](https://rankshieldlegal.com/ai-legal-citation-checker/).
A general model can produce citations that look real and do not exist. Reasonable understanding means knowing that in advance and verifying every cited authority against the source. Do not trust any claim that a tool is hallucination-free.

## Confidentiality is part of competence, and it is not privilege
Knowing where inputs go is part of competence, because you cannot protect confidentiality in a tool you do not understand. The confidentiality duty under Rule 1.6 is distinct from evidentiary privilege, which courts decide separately and which competence does not resolve.
The second verification point, where inputs go, is where the competence duty meets the confidentiality duty. Opinion 512 folds confidentiality implications into the reasonable understanding a lawyer is expected to have, because you cannot protect client information in a tool if you do not know whether the tool retains it or trains on it [[3]](#ref-3). Understanding the tool and protecting confidentiality are not separate projects. The first is what makes the second possible.
It is worth stating plainly that the confidentiality duty here, under Model Rule 1.6, is not the same as evidentiary privilege. Rule 1.6 is the ethical duty that covers all information relating to a representation, whether or not anything is ever litigated. Privilege is a narrower evidentiary protection that a court can find waived by disclosure to a third party, and whether routing client material through an AI tool affects privilege is a separate, unsettled question that courts decide on the facts. Competence and confidentiality do not resolve the privilege question, and this article does not claim they do.
For competence purposes, the point is narrower and clearer. Knowing where inputs go, and keeping confidential material out of tools that retain or train on it, is part of the reasonable understanding the duty asks for. That is a confidentiality discipline you can build into the checklist above. For how the confidentiality duty plays out across a representation, see [our overview of client confidentiality and AI](https://rankshieldlegal.com/client-confidentiality-ai/), and for turning these expectations into firm rules, our [guide to building a law firm AI policy](https://rankshieldlegal.com/law-firm-ai-policy/).

## Competence does not stop at the lawyer who clicks the button
Opinion 512 connects the competence duty to supervision under Rules 5.1 and 5.3, and that connection changes the scope of what a firm has to do [[3]](#ref-3). It is not enough for the individual lawyer running a query to have a reasonable understanding of the tool. The firm has to oversee how associates, paralegals, and other staff use these tools, so that AI use sits inside a policy rather than happening ad hoc across a dozen individual habits.
This is where competence becomes an organizational obligation rather than a personal one. A partner who understands a tool perfectly has still not met the supervision expectation if associates are pasting client material into consumer chatbots without anyone knowing. The reasonable understanding at the top of the firm has to translate into rules, training, and oversight that reach everyone whose work touches these tools, including nonlawyer staff under Rule 5.3.
The practical implication is that the checklist above is not just a personal exercise for each lawyer. It is the basis for a firm-wide standard: which tools are approved, what may and may not go into them, and how outputs are checked, applied consistently and supervised. Vetting the tools themselves is part of that standard, which is why a structured vendor review belongs alongside it. See our [security questionnaire for vetting a legal AI vendor](https://rankshieldlegal.com/vet-legal-ai-vendor-security-questionnaire/) for a way to make that review repeatable.
Under Rules 5.1 and 5.3, competence extends to supervising how everyone in the firm uses AI, including nonlawyer staff. Individual understanding is necessary but not sufficient; the firm needs a supervised standard.

## Where verification tooling helps, and where it does not
There is a gap between meeting the competence duty and being able to show later that you met it. A firm can understand its tools, keep confidential material out of the wrong ones, and check every citation, and still, months on, have only recollection to show for it. The competence obligation is continuous, but the evidence that it was honored tends to evaporate unless something captures it at the time.
This is the narrow thing verification tooling can help with, and it is worth being precise about the claim. RankShield Legal's citation certification can attest that cited authorities in a document were checked against real sources, which is a direct piece of evidence for the output-checking step the competence duty asks for. Privilege-isolation attestation, part of our roadmap rather than a claim of current general availability, is designed to attest that confidential or privileged material was isolated from a tool, whether by being withheld, redacted, tokenized, or handled by a local model. Both attest to process and to architecture. Neither decides a legal outcome.
The honest boundary is the same one this article has held throughout. An attestation can show that citations were verified and that sensitive material was handled as the firm's policy required. It cannot make a lawyer competent, and it cannot decide whether privilege was preserved, because that remains a determination a court makes on the facts. What it changes is not the standard the rules set. It changes whether a firm can demonstrate, rather than assert, that it did the work the duty of competence calls for, at the moment someone asks.
Verification tooling evidences that citations were checked and that material was isolated. Citation certification is available; privilege-isolation attestation is a roadmap item. Neither makes a lawyer competent, and neither decides whether privilege was preserved.

## Questions to work through with your own counsel
None of this has to be reinvented for every tool or every matter. The same short set of questions applies each time, and a firm that answers them before an AI tool touches client work is running the competence analysis by habit rather than by luck. These questions map to the verification points above, and they are meant to structure a conversation with a licensed attorney in your jurisdiction, not to substitute for one.
This article is general information from a founder who builds verification tools for law firms, not legal advice, and not an opinion on any specific matter. Comment 8 has been adopted in varying forms across most states, the confidentiality duty is distinct from privilege, and how these duties apply is fact-specific. Use the questions below as a starting point, then take the actual decision to your own counsel and your jurisdiction's adopted rules.

- Can you state, in plain terms, what each AI tool you use does and does not do, including its tendency to fabricate citations [[3]](#ref-3)?
- Do you know where each tool sends your inputs, and whether they are retained or used for training?
- Does the tool's confidentiality posture match the sensitivity of the information you put into it, under Rule 1.6?
- Is there a concrete step that verifies every cited authority against its source before filing [[3]](#ref-3)?
- Is staff use of these tools supervised under a firm policy, consistent with Rules 5.1 and 5.3 [[3]](#ref-3)?
- Can the firm later show, not just assert, that these steps happened for a given matter?
This is informational, not legal advice. Because technology competence is fact-specific and varies by jurisdiction, consult a licensed attorney in your jurisdiction before relying on any AI workflow.

Test yourself
## Test your technology competence
A short quiz on what Rule 1.1 and Opinion 512 actually require.

- 1 What level of AI understanding does Opinion 512 require of a lawyer? Expert-level machine learning knowledge A reasonable understanding of the specific tools you use No understanding at all **Answer:** A reasonable understanding of the specific tools you use Opinion 512 is explicit that a lawyer need not become an AI expert, but must have a reasonable understanding of the capabilities and limitations of the specific tools in use.
- 2 Which three questions cover most of the competence duty for an AI tool? Price, brand, and popularity What the tool does, where your inputs go, and how outputs are checked Architecture, weights, and source code **Answer:** What the tool does, where your inputs go, and how outputs are checked The duty resolves into these three. What the tool does is usually documented, where inputs go is a contract question, and how outputs are checked is your own process. None requires technical training.
- 3 What is the clearest competence failure with a general large language model? Slow output Fabricated citations that must be verified against the actual source A high subscription cost **Answer:** Fabricated citations that must be verified against the actual source A general model can produce citations that look real but do not exist. Reasonable understanding means knowing that in advance and verifying every cited authority. No tool should be described as free of hallucinations.
- 4 Does the competence duty extend to associates and nonlawyer staff? No, only the lawyer running the query Yes, through supervision under Rules 5.1 and 5.3 Only to partners **Answer:** Yes, through supervision under Rules 5.1 and 5.3 Opinion 512 connects competence to supervision under Rules 5.1 and 5.3, so the firm must translate understanding into rules, training, and oversight that reach everyone whose work touches these tools.
Honest self-check. There is no sign-up, and nothing is stored.

Questions answered
## Straight answers to the common questions
The questions readers ask about this topic, answered directly. **No forms, no sales pitch.**

JAMIE KLONCZ · SEO AGENCY NAPLES ************** ONLINE
Pick a question on the left, or search above. You will get the direct answer, the way an answer engine would give it.

← PREV NEXT → [REQUEST ACCESS →](https://rankshieldlegal.com/contact/)

- **Does the duty of competence really require me to understand AI?** Yes, at a reasonable level, for the specific tools you use. Model Rule 1.1 requires competent representation, and Comment 8 says a lawyer should keep abreast of the benefits and risks associated with relevant technology [1][2]. ABA Formal Opinion 512 applies that to generative AI and sets the bar clearly: a lawyer need not become an AI expert but must have a reasonable understanding of the capabilities and limitations of the specific AI tools the lawyer uses [3]. Most states have adopted a version of Comment 8, so a technology competence expectation very likely applies to your practice. The duty is not abstract. It resolves into knowing what a tool does and does not do, where your inputs go, and how you check outputs, none of which requires technical training. Check your own jurisdiction's adopted rule for the exact wording that governs you.
- **Do I have to become an AI expert to use these tools?** No. Opinion 512 is explicit that a lawyer need not become an AI expert to use these tools competently, and instead sets the standard at a reasonable understanding of the specific tools in use [3]. The distinction matters in practice. You do not need to understand a model's architecture, training mathematics, or source code. You do need to know what the tool produces and does not produce, whether it retains or trains on your inputs, and how you verify its outputs before relying on them. Most of that is answerable from the vendor's documentation and contract read critically, plus your own checking process, rather than from any technical background. Reasonable understanding is a standard a firm can meet with a defined process, not one that requires hiring a data scientist.
- **What exactly am I supposed to verify about an AI tool?** Three things cover most of the duty. First, what the tool does and does not do, including that a general large language model can produce fabricated citations, which is why cited authorities have to be verified rather than trusted [3]. Second, where your inputs go, meaning whether they leave your control, whether they are retained, and whether they are used to train the model, which you confirm from the contract and configuration rather than the marketing page. Third, how outputs are checked, which is your own process for catching errors before they reach a filing or a client. The first is usually documented, the second is a contractual question, and the third is entirely within your control. A firm that can answer all three in plain terms for a tool is close to the reasonable understanding the duty asks for.
- **Is the confidentiality part of this the same as attorney-client privilege?** No, and keeping them separate matters. The confidentiality duty tied to competence here is the ethical duty under Model Rule 1.6, which covers all information relating to a representation, whether or not anything is ever litigated. Opinion 512 folds confidentiality implications into the reasonable understanding a lawyer should have, because you cannot protect client information in a tool without knowing whether it retains or trains on that information [3]. Evidentiary privilege is a narrower protection that a court can find waived by disclosure to a third party, and whether routing client material through an AI tool affects privilege is a separate, unsettled question that courts decide on the facts. Understanding a tool and protecting confidentiality do not resolve the privilege question. Treat privilege as its own analysis, decided by a court, not something competence settles.
- **Does technology competence extend to my associates and staff?** Yes. Opinion 512 connects the competence duty to supervision under Model Rules 5.1 and 5.3, so competence is an organizational obligation, not just a personal one [3]. It is not enough for the lawyer running a query to understand the tool if associates or nonlawyer staff are using AI in ways no one oversees. The firm has to translate that understanding into rules, training, and supervision that reach everyone whose work touches these tools. In practice, that means an approved-tools list, clear limits on what may go into each tool, a consistent output-checking step, and oversight of how staff actually use them. A partner who personally understands a tool has still not met the supervision expectation if the firm's use of it is unmanaged. The competence checklist is the basis for that firm-wide standard, not just a personal exercise.
- **Can a tool prove I met the competence duty?** A tool cannot make a lawyer competent, but it can evidence some of the steps competence requires. RankShield Legal's citation certification can attest that cited authorities were checked against real sources, which is direct evidence for the output-checking step Opinion 512 expects [3]. Privilege-isolation attestation, a roadmap item rather than a claim of current general availability, is designed to attest that confidential material was isolated from a tool. Both attest to process and architecture, not to a legal outcome. No tool is hallucination-free, and none can decide whether privilege was preserved, which remains a court's determination. What honest tooling changes is whether a firm can later demonstrate, rather than assert, that it understood its tools, protected confidentiality, and verified outputs. The competence judgment itself still belongs to the lawyer and, ultimately, to the rules your jurisdiction has adopted.

## References

- American Bar Association. Model Rule 1.1: Competence. 2024. [https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/)
- American Bar Association. Model Rule 1.1 Competence: Comment (Maintaining Competence, Comment 8). 2024. [https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/comment_on_rule_1_1/](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_1_competence/comment_on_rule_1_1/)
- ABA Standing Committee on Ethics and Professional Responsibility. Formal Opinion 512: Generative Artificial Intelligence Tools. July 29, 2024. [https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/](https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/)
- American Bar Association. Model Rule 5.3: Responsibilities Regarding Nonlawyer Assistance. 2024. [https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_5_3_responsibilities_regarding_nonlawyer_assistant/](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_5_3_responsibilities_regarding_nonlawyer_assistant/)
- American Bar Association. Model Rule 1.6: Confidentiality of Information. 2024. [https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_6_confidentiality_of_information/](https://www.americanbar.org/groups/professional_responsibility/publications/model_rules_of_professional_conduct/rule_1_6_confidentiality_of_information/)

Written by
## [Jamie Kloncz](https://rankshieldlegal.com/about/)
Founder, RankShield
Jamie Kloncz is the founder of RankShield, the verifiable AI and quantum security platform behind RankShield Legal. An engineer by training, he built RankShield after his own devices and business were attacked, including an AI voice-cloning scam that targeted his family, on one conviction: unverifiable security is the real danger, so every consequential action should leave a receipt anyone can independently check.
[More about Jamie →](https://rankshieldlegal.com/about/)

Try it · Free
## Check a citation against live case-law
Paste a citation from an AI-drafted brief and see whether the case actually exists, resolved against live case-law. Free, no sign-up. Then request early access to certify a full filing.
[Try the citation checker](https://rankshieldlegal.com/ai-legal-citation-checker/)

Keep reading
## Related guides
[Legal AI When Your AI Stops Drafting and Starts Acting: The Agentic AI Governance Gap in Law Firms Read guide →](https://rankshieldlegal.com/blog/agentic-ai-governance-gap-law-firms/)[Legal AI California Is Turning AI Ethics Guidance Into Disciplinary Rules Read guide →](https://rankshieldlegal.com/blog/california-ai-disciplinary-rules/)[Legal AI What Court AI-Certification Orders Require You to Attest Read guide →](https://rankshieldlegal.com/blog/court-ai-certification-orders/)