Step 4: Review

Hold paramount the safety, health, and welfare of the public.

Perform services only in areas of their competence.

Avoid deceptive acts.

Engineers shall not reveal facts, data, or information without the prior consent of the client or employer except as authorized or required by law or this Code.

Engineers shall undertake assignments only when qualified by education or experience in the specific technical fields involved.

Engineers shall not affix their signatures to any plans or documents dealing with subject matter in which they lack competence, nor to any plan or document not prepared under their direction and control.

Engineers shall avoid all conduct or practice that deceives the public.

Engineers shall conform with state registration laws in the practice of engineering.

Engineers shall give credit for engineering work to those to whom credit is due, and will recognize the proprietary interests of others.

The Board cited this case as a historical parallel establishing that new drafting/design technology (CADD) can ethically be used and sealed by an engineer, provided the engineer maintains competency and does not use it as a substitute for judgment, foreshadowing the analysis of AI use.

The Board cited this case to address the omission of key safety features and lack of competency oversight in AI-generated design documents, using it to show that technology must never replace engineering judgment, while distinguishing Engineer A's overall competence from the incompetent engineer in that case.

Engineer A's use of AI in report writing was partly ethical, and partly unethical. Engineer A was competent and did thoroughly review and verify the AI-generated content, ensuring accuracy and compliance with professional standards. However, Engineer A did not obtain client permission to disclose private information, nor did Engineer A document required technical citations. Ethical use of AI to create the report text must satisfy all pertinent requirements.

The use of AI-assisted drafting tools by Engineer A was not unethical per se. However, Engineer A’s misuse of the tool, by failing to maintain Responsible Charge over the AI tool and its output before sealing the document and providing it to Client W, was unethical.

Similar to other software used in the design or detailing process, Engineer A has no professional or ethical obligation to disclose AI use to Client W (unless such disclosure is required under Engineer A’s contract with Client W). However, at the time of the BER’s review of this case there is no universal guideline mandating AI disclosure in engineering work. Ethical principles favor transparency when AI plays a substantial role in generating work products. To uphold ethical standards, engineers integrating AI into their practice should adopt rigorous verification processes and consider disclosing AI involvement when it plays a significant role in the final product.

The Board's finding that Engineer A failed to obtain client permission to disclose private information can be extended: the confidentiality violation crystallized at the moment Client W's groundwater and site data were entered into the open-source AI platform, independent of whether the resulting text was later verified or accurate. Thorough post-hoc review of AI output cannot retroactively cure an unauthorized disclosure that occurs upstream in the data-input stage, particularly where the platform's data retention and reuse practices are unknown to the engineer. This suggests the ethical breach under II.1.c is best understood as a discrete act of disclosure, not a data-quality issue remediable by verification.

The Board's responsible-charge finding can be deepened by noting an internal inconsistency in Engineer A's conduct: applying rigorous, multi-source verification to the report while giving only a cursory review to the sealed design documents suggests Engineer A implicitly (and incorrectly) treated the two AI outputs as carrying different risk profiles, when in fact the design documents—bearing Engineer A's seal and governing physical infrastructure with public safety implications—warranted at least equivalent, if not greater, scrutiny than the report. This disparity indicates that the lapse in responsible charge was not merely a matter of insufficient time or diligence but reflected a misjudgment about where professional risk was concentrated, undermining the presumption that professional judgment alone can substitute for adequate review calibrated to consequence severity.

Beyond the Board's transparency recommendation, a distinct obligation exists that does not depend on resolving the broader AI-disclosure debate: under III.9, engineers must give credit for engineering work to those to whom credit is due. Because the AI-generated introduction was polished enough that Client W perceived two different authorial voices, this suggests Engineer A's failure to cite the AI tool was not merely a matter of optional transparency but potentially implicated a distinct intellectual honesty and credit-attribution duty, separate from confidentiality or competence concerns, that exists independent of whether industry-wide AI disclosure norms have yet solidified.

The Board's analysis does not directly address whether Engineer A's loss of mentor-based quality assurance created an independent obligation to secure alternative human review before relying on an unfamiliar AI tool as a substitute for that function. Since AI tools cannot exercise professional judgment or bear responsibility, the disappearance of Engineer B's review capacity arguably heightened, rather than diminished, Engineer A's duty to seek some form of qualified peer or supervisory review, especially for sealed design documents; using an untested AI tool as a functional replacement for human mentorship, rather than as a supplement to it, may itself represent a distinct lapse in professional judgment separate from the review-adequacy failure the Board identified.

Q101: The unauthorized disclosure occurred at the moment Engineer A entered Client W's confidential groundwater and site data into the open-source AI platform, independent of whether the AI-generated text was later verified for accuracy. Confidentiality under II.1.c is a duty tied to the act of disclosure itself; subsequent verification of the resulting text addresses accuracy and originality concerns but does nothing to cure the prior unauthorized transmission of client data to a third-party system of unknown retention and use practices.

Q102: The disparity between Engineer A's thorough review of the report and cursory review of the sealed design documents suggests a misapplication of risk-based judgment. Sealed engineering plans carry direct public safety consequences and trigger the heightened obligations of II.2.b and I.1, whereas a draft report explicitly labeled as such carries comparatively lower immediate risk. Engineer A's failure to allocate greater scrutiny to the higher-risk work product indicates an inconsistent, rather than risk-calibrated, approach to reviewing AI output.

Q103/Q104: Beyond any general AI-disclosure question, Engineer A's competence obligations under I.2 and II.2.a arguably required flagging to Client W that the drafting tool was new to market and entirely unfamiliar, and required Engineer A to seek an alternative qualified human reviewer once Engineer B retired. Substituting an untested AI tool for the lost mentorship function, without securing comparable independent review, left a quality-assurance gap that Engineer A's own competence duty should have prompted them to fill.

Q201: The tension between confidentiality and efficiency should be resolved in favor of confidentiality as a threshold constraint, not a factor to be balanced against convenience. Where an AI tool's data-handling practices are unknown, engineers should either obtain informed client consent before inputting proprietary data, or use de-identified/anonymized data sufficient to preserve analytical utility without exposing protected information.

Q202/Q204: Yes, a genuine conflict exists. Engineer A's own professional judgment determined that a high-level review of the design documents was sufficient, yet that same judgment proved miscalibrated when the AI output later revealed misaligned dimensions and omitted safety features. This indicates that 'professional judgment primacy' cannot be self-certifying; it must be anchored to objective, risk-proportionate review protocols, especially for documents destined for sealing under II.2.b.

Q203: Even absent a formal citation standard for AI-generated text, the fact that Client W perceived the report as written by two different authors is itself evidence that the AI's contribution was substantial enough to trigger the spirit of III.9's credit-attribution principle. Intellectual honesty does not depend on the existence of a specific rule; the stylistic discontinuity noticed by the client demonstrates that failure to disclose AI involvement created a misleading impression of authorship.

Q301: From a strict deontological reading of II.1.c, Engineer A did not fulfill the confidentiality duty. The duty not to reveal client facts or data without prior consent is unconditional and act-based; it is breached the moment the data is transmitted to an outside system without consent, regardless of the AI tool's actual security or the client's subsequent satisfaction with the resulting work product.

Q302: A consequentialist framing focused solely on Client W's satisfaction is insufficient to justify Engineer A's undisclosed and initially unverified AI drafting. Consequentialist analysis in a professional ethics context must account for broader systemic harms beyond a single client's experience, including risks to public trust in engineering authorship, potential downstream reliance by third parties on unattributed AI content, and precedent effects on the profession's credibility.

Q303: Viewed through a virtue-ethics lens, Engineer A's cursory review before sealing the design documents reflects a deficiency in professional prudence and diligence rather than a deliberate ethical lapse. The act of sealing is a public trust gesture that calls for the virtue of thoroughness commensurate with the gravity of the representation being made; a rushed review, even if well-intentioned, falls short of the integrity the sealing act demands.

Q304: Under a deontological view of responsible charge, Engineer A did not satisfy the duty because responsible charge requires the sealing engineer to have personally and substantively verified the engineering content prior to sealing, not merely to have made superficial site-specific adjustments to AI output from an unfamiliar tool. The duty is process-oriented and non-delegable, meaning the AI's role as a drafting aid does not reduce Engineer A's independent verification obligation.

Q401: Had Engineer A not cross-checked the AI-generated report against professional journal articles and search-engine queries, the Board would very likely have found the report-writing conduct wholly unethical rather than partly ethical. The thorough verification was the specific factor that satisfied the accuracy and competence elements of the Board's analysis; without it, the report would have suffered from unverified accuracy in addition to the confidentiality and citation deficiencies already found.

Q402: Even if Engineer B had remained available for mentorship, the Board would likely still have found a responsible charge lapse, because the duty to maintain responsible charge over sealed documents rests personally with the sealing engineer and cannot be fully discharged through a mentor's informal review. Availability of Engineer B might have reduced the likelihood of errors reaching the client, but it would not substitute for Engineer A's own obligation to verify AI output before sealing.

Q403: The Board's conclusion regarding misuse of the AI tool and lapse in responsible charge rests on the adequacy of Engineer A's review process, not on the discovery of actual errors. Even if Client W had not caught the misaligned dimensions and omitted safety features, the cursory review itself would still constitute a failure to maintain responsible charge, since the standard concerns the diligence exercised before sealing rather than whether errors happen to surface afterward.

Q404: The confidentiality problem identified by the Board is specifically tied to the act of inputting Client W's proprietary groundwater and site data into the open-source AI tool. Had Engineer A instead used only publicly available or non-proprietary information to draft the report, no confidentiality violation under II.1.c would have arisen from the AI use itself, though the citation and disclosure concerns regarding AI-generated text would remain unaffected.

The case reveals that Review Adequacy of AI Output is not a fixed standard but must scale with the risk and finality of the work product. Engineer A applied rigorous, multi-source verification to the report text but only a cursory review to the sealed design documents, even though both originated from the same unfamiliar AI tool. This asymmetry shows that Responsible Charge over AI-Assisted Design cannot be satisfied by a uniform review effort; instead, the level of scrutiny must be calibrated to the document's downstream consequences (a revisable draft versus a sealed engineering document that directly implicates public safety). Where Engineer A treated both outputs with equivalent methodological trust despite their unequal stakes, the principle of Public Welfare in AI Design Errors was subordinated to convenience, and the tension was not resolved but simply avoided until Client W's discovery of the deficiencies forced it into the open.

Confidentiality of Client W Information and the practical efficiency of using an open-source AI tool were never actually balanced, because Engineer A never paused to weigh them: client data was input into a third-party platform of unknown retention and training practices without first obtaining consent or assessing the tool's data-handling policies. This suggests that in AI-assisted practice, confidentiality obligations under II.1.c. function as a threshold gate that must be cleared before efficiency considerations are even relevant, not as a factor to be traded off against them. Because Engineer A's technical competence and thorough fact-checking could not cure the upstream disclosure, the case demonstrates that procedural competence in verifying AI output is a separate ethical dimension from the prior, non-negotiable duty to protect client information before it is ever shared with an AI system.

The Board's treatment of Transparency of AI Contribution as aspirational rather than mandatory, absent a governing citation standard, illustrates a broader prioritization principle: codified duties (competence, confidentiality, responsible charge) take precedence over emerging normative expectations (AI disclosure) until the profession formalizes the latter into enforceable rules. Yet the Board's simultaneous recommendation that engineers 'consider disclosing' AI involvement shows that Intellectual Honesty in AI Text and Credit Attribution in AI Report are treated as values in tension with, but not yet enforceable against, the current Code text (III.9., I.5.). This gap is precisely what allowed Engineer A's uncited use of AI to remain technically compliant even though it produced a report so stylistically inconsistent that Client W perceived two different authors—showing that formal compliance and substantive intellectual honesty can diverge when governing standards lag behind technological practice.

Was Engineer A’s use of AI to create the report text ethical, given that Engineer A thoroughly checked the report?

Was Engineer A’s use of AI-assisted drafting tools to create the engineering design documents ethical, given that Engineer A reviewed the design at a high level?

If the use of AI was acceptable, did Engineer A have an ethical obligation to disclose the use of AI in any form to the Client?

Did Engineer A's act of inputting Client W's confidential groundwater and site data into an open-source AI platform constitute an unauthorized disclosure of client information, regardless of whether the AI-generated text was later verified?

Why did Engineer A apply thorough verification to the report but only a cursory review to the engineering design documents, and does this disparity suggest a failure to recognize that sealed design documents carry greater public safety risk than a draft report?

Should Engineer A have disclosed to Client W that the AI drafting tool being used for both the report and the design documents was new to the market and entirely unfamiliar to Engineer A, independent of any later disclosure of AI use generally?

Given that Engineer B's retirement left Engineer A without a mentor for quality assurance review, did Engineer A have an obligation to secure an alternative qualified human reviewer before relying on an unfamiliar AI tool as a substitute for that review function?

How should the Confidentiality of Client W Information be balanced against the practical efficiency gains of using open-source AI tools that require inputting client data into third-party systems of unknown data retention practices?

Does Review Adequacy of AI Output conflict with Responsible Charge over AI-Assisted Design when Engineer A applied rigorous verification to the report but only a high-level review to the design documents, despite both originating from the same unfamiliar AI tool?

How should Transparency of AI Contribution be balanced against Intellectual Honesty in AI Text and Credit Attribution in AI Report when no professional citation standard currently mandates disclosure, yet the AI-generated introduction was polished enough that Client W perceived the report as written by two different authors?

How should Professional Judgment Primacy in AI Use be reconciled with Public Welfare in AI Design Errors, given that Engineer A's own professional judgment approved a cursory review of design documents that later proved to omit required safety features?

From a deontological perspective, did Engineer A fulfill the duty of confidentiality under II.1.c by inputting Client W's proprietary information into an open-source AI tool without first obtaining consent?

From a consequentialist perspective, does the fact that Client W ultimately found the draft report satisfactory justify Engineer A's undisclosed and unverified use of AI in drafting the introductory section?

Did Engineer A act with professional integrity, in the virtue-ethical sense, when conducting only a cursory review of the AI-generated design documents before affixing their professional seal?

From a deontological perspective, did Engineer A satisfy the duty of maintaining responsible charge required before sealing engineering documents whose substantive content was generated by an unfamiliar AI tool?

If Engineer A had not thoroughly cross-checked the AI-generated report against professional journal articles and search-engine queries, would the Board still have found the report-writing conduct partly ethical rather than wholly unethical?

If Engineer B had remained available to provide mentorship and quality-assurance review, would the Board still have concluded that Engineer A failed to maintain responsible charge over the AI-generated design documents?

If Client W had not discovered the misaligned dimensions and omitted safety features in the AI-generated design documents, would the Board still have concluded that Engineer A's cursory review constituted a misuse of the AI tool and a lapse in responsible charge?

If Engineer A had not input Client W's information into the open-source AI software, would the Board still have found a confidentiality problem in Engineer A's use of AI to draft the report?

Because AI Tool Adoption was guided by the obligation to perform services only in areas of competence yet proceeded without adequately verifying that competence boundary, it set in motion the chain leading to Confidential Data Input and the eventual exposure of protected information.

Revision Instruction is guided by the duty to hold paramount public safety, health, and welfare, which matters because it is the corrective response triggered by the Design Error Discovery that arose from an inadequately reviewed AI-generated design, showing the obligation functioning as a safeguard against propagating a flawed design into practice.

Confidential Data Input violates Client Confidentiality precisely because it directly causes Confidential Information Exposure, illustrating how the convenience of using the AI tool was purchased at the cost of a core duty to protect client data.

Thorough Report Review fulfills the Direction and Control obligation and is guided by the duties to avoid deceptive acts and to work only within competence, which matters because this review was the intended checkpoint for catching AI-introduced errors before the report was sealed and submitted, yet its outcome still allowed problems to surface later as a Report Inconsistency Observation.

Report Sealing and Submission violates the duty to Give Credit for Engineering Work because sealing the report as the engineer's own product while omitting the AI tool's role directly precipitates the later Report Inconsistency Observation, exposing the misrepresentation embedded in the submission.

Because Engineer A's cursory review of AI-generated designs directly caused the Design Error Discovery that later forced Client W's Revision Instruction, the failure to exercise the direction and control and responsible charge required by the Code shows how skipping careful human oversight of AI output let a flawed design nearly proceed unchecked, undermining the paramount duty to protect public safety and welfare.

Since this omission sits alongside the AI-driven design and reporting process that produced the inconsistencies Engineer A later noticed, failing to disclose the AI tool's role denies proper credit for the work actually performed and obscures the true authorship and reliability of outputs feeding into decisions like the design revision.

The question arises because Engineer A's thorough review satisfies competence and accuracy obligations, yet this same act of using AI-generated text without disclosure remains contested against separate honesty and credit-attribution norms, leaving the ethical status unresolved.

The question arises because Engineer A substituted a cursory, high-level check for the kind of independent, detailed review normally expected of the engineer in responsible charge, and a subsequent design error discovery exposed the gap between the level of review performed and the level required by the public welfare and responsible charge warrants.

The question arose because Engineer A's undisclosed use of AI in report and design work forced a comparison between traditional credit attribution norms for human collaborators and the ambiguous status of AI as either a tool or a contributing author, with no clear rule (Absent AI Disclosure Guidelines) resolving which warrant governs.

The question emerged because the physical act of data disclosure and the subsequent verification of AI output are treated as separable events under different warrants, forcing scrutiny of whether harm requires actual misuse or is constituted by the unauthorized transfer itself.

The question arises because Engineer A's differing levels of diligence across two AI-generated deliverables expose an unresolved prioritization between competence-based review duties and the higher stakes of sealed public safety documents, creating ambiguity about whether professional judgment was properly calibrated to risk.

The question arises because Engineer A's use of an untested, unfamiliar AI tool created a gap between what the client was told (general AI use) and what the client may have needed to know (the tool's complete novelty), leaving it unclear whether competence and honesty obligations extend beyond disclosing AI use itself to disclosing the engineer's own unfamiliarity with the specific tool.

The question arose because the loss of an established mentorship based QA process coincided with adoption of a new and unverified technology, creating ambiguity about whether procedural safeguards (a human reviewer) were mandatory or whether outcome based safeguards (adequate self review) could satisfy the same underlying duty.

The question arises because Engineer A's data input action satisfies an efficiency-oriented professional practice norm while simultaneously violating the explicit confidentiality obligation, and the lack of clarity about the third party tool's data retention practices leaves the applicability of each warrant genuinely contested.

The question arises because identical AI unfamiliarity produced inconsistent review rigor across two deliverables from the same engineer, exposing an internal inconsistency in how the responsible charge and review adequacy obligations were discharged.

The question arose because Engineer A used an AI tool to draft polished report text without informing Client W, and the resulting stylistic mismatch caused Client W to independently detect and question authorship, revealing a gap between formal citation rules and implicit expectations of honest attribution.

The question arises because the same act, Engineer A's approval of a cursory review, is the evidence used both to invoke professional autonomy and to indict that autonomy for failing the public welfare obligation, leaving no clear rule for how much deference judgment deserves when its output is demonstrably unsafe.

The question emerged because Engineer A's unilateral act of feeding Client W's confidential information into an AI system created an unresolved conflict between a strict duty-based reading of the confidentiality clause and a more permissive reading grounded in professional discretion over tool selection, with no clear precedent (e.g., BER Case 90-6 or 98-3) settling whether AI input equates to disclosure.

The question arose because the case data separates process (undisclosed AI drafting) from outcome (client satisfaction), forcing a choice between judging the action by its result or by whether it honored disclosure and honesty obligations regardless of result.

The question arises because sealing a design with only cursory review creates a visible gap between the professional's certifying authority and the actual diligence performed, and it is unresolved whether virtue ethics judges this gap by the outcome, the intent, or the habitual disposition it reveals.

The question arises because sealing engineering documents traditionally presumes the sealer possesses full command of the underlying content, but Engineer A's reliance on an unfamiliar AI tool combined with only a cursory design review breaks that presumption and forces a choice between deontological duties that the facts do not clearly resolve.

The question arises because the Board's partial ethicality finding rests on an empirical fact, the cross-checking behavior, that could have been otherwise, exposing that the ethical verdict is contingent on satisfying one warrant (review duty) even while violating another (citation duty).

The question arises because Engineer B's retirement removed a review mechanism at the same time Engineer A committed a cursory review of AI output, creating ambiguity about whether the Board's finding of a responsible-charge failure depended on that missing support structure or would hold independent of it.

The question arises because the Board's actual finding depended on facts discovered after Engineer A's review (the omitted safety features and dimensional errors), raising doubt about whether the violation was judged on the review process itself or on the negative outcome that was nearly realized.

This question arose because the case record bundles the confidentiality breach (Confidential Data Input) and the AI drafting practice (AI Report Generation) into one factual event, leaving it unclear whether the Board's ethical objection was to the data disclosure specifically or to unsupervised AI use generally.

Given that Engineer A verified the AI-generated report's accuracy but separately failed to secure client consent for disclosed data and omitted citations, the board concluded the conduct was partly ethical and partly unethical, treating competence and consent/citation duties as independently assessable requirements.

Given that Engineer A relied on an unfamiliar AI tool for sealed design documents but reviewed them only at a high level without replacing the lost mentor safeguard, the board found the AI use itself permissible while the resulting lapse in responsible charge was the ethical violation.

Given that no universal standard required AI disclosure and Engineer A's tool use resembled ordinary software assistance, the board found no strict ethical obligation to disclose, while still urging transparency as good practice given the AI's substantial role in the polished report.

Given that Engineer A uploaded Client W's confidential data to an open platform of unknown retention practices without consent, the board concluded the confidentiality breach occurred at the point of input, unaffected by any later verification of the AI output's accuracy.

Given that Engineer A devoted greater verification effort to the report than to the higher-stakes sealed design documents from the same unfamiliar AI tool, the board inferred a misjudgment of relative risk, concluding the responsible-charge lapse reflected substantive miscalibration rather than mere insufficient time.

Given that Client W could perceive a shift in authorial voice from the AI-drafted introduction, the board concluded that III.9's credit-attribution duty was implicated independent of whether broader AI-disclosure norms have solidified.

Because Engineer B's retirement removed the customary human check on Engineer A's work, the board reasoned that reliance on an unfamiliar AI tool as a substitute for that function, rather than a supplement to it, represented a separate lapse in judgment.

Because Engineer A uploaded Client W's confidential site data to an open-source AI platform without consent, the board concluded the confidentiality violation occurred instantly and independent of any later verification of the AI output.

Given that Engineer A devoted rigorous review to the lower-risk draft report but only a high-level check to the higher-risk sealed design documents, the board concluded this disparity reflected inconsistent rather than risk-calibrated judgment.

Because the drafting tool was untested and Engineer A lacked mentor support after Engineer B's retirement, the board concluded that competence obligations required disclosure of the tool's novelty to Client W and pursuit of an alternative reviewer, neither of which occurred.

Because Engineer A could not verify how the open-source AI platform would store or reuse Client W's proprietary groundwater data, and no consent was sought beforehand, the board concluded that efficiency gains cannot override the prior duty to protect client information, making consent or de-identification mandatory preconditions rather than optional best practices.

Given that Engineer A's own judgment deemed a cursory review sufficient yet that same review failed to catch misaligned dimensions and missing safety features, the board found a genuine conflict showing that professional judgment must be anchored to external, risk-proportionate protocols rather than trusted purely on the engineer's say-so.

Even though no professional citation standard specifically addressed AI-generated text, Client W's observation that the report read as though written by two different people was itself evidence that the undisclosed AI contribution was substantial enough to violate the spirit of credit-attribution obligations.

Applying a strict deontological lens, the board concluded that the duty not to reveal client data without consent was violated the instant Engineer A input Client W's information into the AI tool, since this duty is act-based and does not depend on whether the tool was later shown to be secure or whether the client was satisfied.

Because Client W's satisfaction was the only outcome considered in the narrow consequentialist argument, the board found this insufficient, reasoning that undisclosed AI drafting risks broader harms such as eroded public trust and unattributed content being relied upon by third parties, harms that a single satisfied client cannot offset.

Given that Engineer A's review was only cursory and the sealed documents came from an unfamiliar AI tool, the board concluded this reflects a deficiency in the virtue of professional thoroughness rather than intentional wrongdoing, because the seal itself demands diligence proportionate to the public trust it conveys.

Because Engineer A only made superficial adjustments to unfamiliar AI output rather than substantively verifying it, the board concluded the deontological duty of responsible charge was not met, since that duty is non-delegable and process-oriented regardless of how the drafting was produced.

Because Engineer A actually cross-checked the AI-generated report against journal articles and search queries, the board concluded this verification specifically cured the accuracy and competence concerns, and absent it the report would have failed on those grounds in addition to its confidentiality and citation problems.

Even hypothesizing that Engineer B remained available, the board concluded a responsible charge lapse would still be found, because informal mentor review does not substitute for Engineer A's own personal obligation to verify AI output before sealing.

Because the board's evaluation rests on the adequacy of Engineer A's review process rather than on whether Client W happened to catch the errors, it concluded that the misuse of the AI tool and the responsible charge lapse would still be found even if the misaligned dimensions and omitted safety features had gone undiscovered.

Given that the confidentiality violation was traced specifically to the proprietary content Engineer A uploaded, the board concluded that substituting public data would have removed the confidentiality problem, while leaving the separate, unresolved question of AI-text disclosure untouched.

Because Engineer A trusted the same unfamiliar AI tool equally despite the design documents carrying far greater public safety consequences than the draft report, the board concluded that review adequacy must scale with risk, and that this uncalibrated scrutiny left the tension between convenience and public welfare unresolved until Client W's discovery forced it into view.

Because Engineer A never paused to assess the AI tool's data practices or secure client consent before inputting proprietary information, the board concluded that confidentiality operates as a non-negotiable gate preceding any efficiency tradeoff, so the subsequent careful verification of AI output could not cure the prior disclosure.

Given that the current Code contains no rule requiring AI disclosure, the board concluded that Engineer A's uncited use remained technically compliant even though it produced a stylistic inconsistency perceptible to Client W, revealing a gap between formal compliance and substantive intellectual honesty that only future codification could close.

Should Engineer A conduct a full, risk-calibrated independent technical review of the AI-generated design documents before sealing them, or rely on a high-level (cursory) review of the AI output?

Should Engineer A disclose the AI's substantial contribution and cite the AI tool and pertinent technical authorities to Client W, or treat the AI output as ordinary drafting software requiring no citation?

Should Engineer A obtain Client W's consent, use de-identified data, or input Client W's full confidential data into the open-source AI platform without consent?

Should Engineer A obtain client consent or de-identify data before inputting Client W's confidential groundwater and site data into an open-source AI platform, or proceed to input the raw data without consent for efficiency?

Should Engineer A apply rigorous, risk-proportionate verification (or secure an alternative qualified reviewer) to the AI-generated design documents before sealing, or rely on a cursory high-level review consistent with the effort applied to the report?

Should Engineer A explicitly cite and credit the AI's substantial contribution to the report text, or treat the AI drafting tool as an uncited internal tool akin to ordinary software?