These answers draw in part from “Ethical Implications Use Of Generative Ai” (CASP CEU Center), and extend it with peer-reviewed research from our library of 27,900+ ABA research articles. Clinical framing, BACB ethics code references, and cross-links below are synthesized by Behaviorist Book Club.
View the original presentation →The Code does not address generative AI by name, but several provisions apply directly. Code 2.01 requires BCBAs to maintain competence in areas relevant to their work—meaning that BCBAs using AI tools are responsible for understanding how these tools function and fail. Code 2.04 requires documentation to accurately represent services.
Code 2.07 requires protection of confidential information—meaning that submitting client-identifiable data to external AI platforms without HIPAA-compliant data processing agreements constitutes a potential ethics violation that practitioners are actively creating, often without realizing it. The most commonly underestimated Code implication is Code 2.04's documentation accuracy requirement: if you sign a document that AI generated and that you did not review with sufficient rigor, you are attesting to its accuracy, and you are responsible for any inaccuracies it contains.
Using AI to generate a first draft is defensible if the resulting document is reviewed with sufficient clinical rigor. Addressing AI-assisted FCT documentation, Dawson et al. (2026) found that specific procedural features of FCT and mand teaching—prompt type, reinforcement schedules, fading procedures—significantly affect outcomes.
BCBAs reviewing AI-generated BIP language should verify that behavioral definitions meet operationalization standards, that objectives trace to functional assessment data, and that recommended procedures match empirically supported approaches for the client's specific behavioral profile. The key question is whether the BCBA's review is genuinely critical—catching the AI's inevitable imprecisions and hallucinations—or merely cursory, providing documentation cover without providing clinical quality assurance.
The fundamental rule is: do not input identifiable client information into an AI platform unless you have verified that the platform operates under a HIPAA Business Associate Agreement. Most consumer-facing AI platforms—including major chatbots—do not meet this standard. BCBAs who want to use AI for clinical documentation should either use healthcare-specific tools designed with HIPAA compliance in mind, or ensure that all identifying information is removed and replaced with generic placeholders before submitting content for AI processing.
BCBAs working in organizations should also ensure that any cloud-based AI tools used by their organization are covered by an organizational-level data use agreement, not only by individual practitioner decisions. Addressing the measurement question directly, Thomas et al. (2026) found that structured feedback procedures produce more reliable behavior change than informal approaches, a finding that applies equally to AI output review — systematic verification consistently outperforms informal spot-checking.
Submitting documentation to insurance carries the practitioner's implicit representation that the document accurately reflects clinical services provided. If AI-generated content has not been thoroughly reviewed and edited by the responsible BCBA, this representation may be false—creating potential billing compliance issues that go beyond BACB Ethics Code violations. BCBAs should apply the same level of review to AI-assisted documentation that they would apply to student-written documentation before signing off: read it critically, correct every inaccuracy, and ensure it reflects your clinical judgment, not a language model's output.
This standard applies equally to AI-generated documentation: the practitioner who submits documentation they did not author and did not rigorously review has misrepresented both the service and their own professional involvement in producing the record.
A valid behavioral definition must identify the behavior in terms that two independent observers would use to record identical data. Test AI-generated definitions by asking: could two different RBTs reading this independently agree on whether to record a tally after observing the same behavioral event? Van & Kubina (2026) reviewed precision measurement approaches and found that clear operational definitions are the prerequisite for reliable behavioral data.
AI-generated definitions that use evaluative language—'engages in appropriate play' rather than 'picks up a toy and manipulates it for at least 3 consecutive seconds'—fail this standard and require rewriting.
This question is not yet settled in law or Code guidance, but the ethical case for proactive disclosure is strong. Families trust BCBAs with sensitive decisions about their children; many would reasonably want to know if AI tools contribute to documents affecting treatment. Proactive disclosure—explaining how AI is used and how the BCBA's judgment governs all outputs—demonstrates the transparency that Code 1.04 models.
Where disclosure might generate disproportionate concern, explaining the BCBA's review process can contextualize AI involvement appropriately. The disclosure conversation also provides an opportunity to explain the BCBA's role as the clinical authority over all outputs, which may actually increase family confidence in service quality rather than reducing it.
AI-generated parent training materials carry two primary risks: technical inaccuracy and misalignment with the specific client's treatment plan. With relevance to AI-assisted remote monitoring, Pichardo et al. (2026) found that caregiver data accuracy depends on the precision of operational definitions they receive—which means imprecise training materials have direct consequences for data quality.
Clinically, generic AI-generated materials may describe procedures that don't match the individualized plan, creating caregiver confusion or protocol inconsistency. The review process for AI-generated parent training materials should specifically check each procedure description against the client's actual individualized treatment plan, since generic materials may describe procedures that have not been authorized for this client.
This is a scope-of-competence and documentation integrity issue simultaneously. The BCBA should raise the concern explicitly: using AI-generated documentation without adequate review creates risk of inaccurate clinical records, potential billing compliance issues, and BACB Ethics Code violations. If the employer does not respond adequately, the BCBA faces a potential conflict between organizational policy and professional obligation.
Code 1.07 applies: if the employer's AI documentation practices constitute an ethics violation, the BCBA has an obligation to address it through available channels. BCBAs who receive this kind of organizational pressure should also consider whether their organization's AI documentation policy is creating systemic ethics risk that warrants formal documentation under Code 1.07.
AI tools are more defensible for literature search assistance than for clinical documentation, but they require the same critical review. Language models hallucinate citations—generating plausible-seeming reference strings that do not correspond to actual publications. Any AI-generated citation must be verified against the actual source before use.
Chang (2026) illustrated how even published literature can mischaracterize ABA practices when treatment descriptions are imprecise—a risk compounded when AI summarizes that literature with further imprecision. BCBAs should read primary sources directly rather than relying on AI-generated summaries.
Formal guidance from the BACB and professional organizations will likely clarify permitted and prohibited uses, but the pace of technology change makes specific rule-following an insufficient long-term strategy. The more durable approach is to maintain a principled framework: AI use is appropriate when it accelerates work within the practitioner's competence, does not substitute for clinical judgment, protects client confidentiality, and produces outputs that meet professional documentation standards after review. Relevant to AI documentation standards, Kok et al.
(2026) found implementation precision determines treatment outcomes—a finding that anchors why documentation quality, however it is produced, cannot be compromised.
The ABA Clubhouse has 60+ on-demand CEUs including ethics, supervision, and clinical topics like this one. Plus a new live CEU every Wednesday.
Ready to go deeper? This course covers this topic with structured learning objectives and CEU credit.
Ethical Implications Use Of Generative Ai — CASP CEU Center · 1 BACB Ethics CEUs · $
Take This Course →We extended these answers with research from our library — dig into the peer-reviewed studies behind the topic, in plain-English summaries written for BCBAs.
279 research articles with practitioner takeaways
252 research articles with practitioner takeaways
239 research articles with practitioner takeaways
You earn CEUs from a dozen different places. Upload any certificate — from here, your employer, conferences, wherever — and always know exactly where you stand. Learning, Ethics, Supervision, all handled.
No credit card required. Cancel anytime.
All behavior-analytic intervention is individualized. The information on this page is for educational purposes and does not constitute clinical advice. Treatment decisions should be informed by the best available published research, individualized assessment, and obtained with the informed consent of the client or their legal guardian. Behavior analysts are responsible for practicing within the boundaries of their competence and adhering to the BACB Ethics Code for Behavior Analysts.