Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health

Beatriz Garcia Santa Cruz; Carlos Vega; Philip Santangelo; Venkata Satagopam

doi:10.1007/978-3-032-06004-4_8

Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health

Beatriz Garcia Santa Cruz^*
, Carlos Vega
, Philip Santangelo
, Venkata Satagopam^*

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

The rapid rise of agentic AI in mental health exposes a misalignment between conversational fluency and user safety. We present a multidimensional framework spanning nine evaluation domains—from clinical validity to relational risk and regulatory compliance—designed to assess agentic systems beyond surface metrics. Applying it to eight real-world tools reveals that the most popular agents often lack basic safeguards, while clinically validated solutions remain underutilized. These findings highlight the need for structured oversight and public awareness. Our framework offers a foundation for safer deployment, informed regulation, and design practices grounded in mental health ethics.

Original language	English
Title of host publication	AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings
Editors	Jianing Qiu, Jinlin Wu, Curtis Langlotz, Baoru Huang, Zhen Lei, Honghan Wu, Hongbin Liu, Weidi Xie
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	74-84
Number of pages	11
ISBN (Print)	9783032060037
DOIs	https://doi.org/10.1007/978-3-032-06004-4_8
Publication status	Published - 22 Sept 2025
Event	1st International Workshop on Agentic AI for Medicine, Agentic AI 2025, 1st International Workshop on Clinical-Driven Robotics and Embodied AI Technology, CREATE 2025, and 1st International Workshop on Multimodal Large Language Models in Clinical Practice, CMLLMs 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025 - Daejeon, Korea, Republic of Duration: 23 Sept 2025 → 27 Sept 2025

Publication series

Name	Lecture Notes in Computer Science
Volume	16147 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	1st International Workshop on Agentic AI for Medicine, Agentic AI 2025, 1st International Workshop on Clinical-Driven Robotics and Embodied AI Technology, CREATE 2025, and 1st International Workshop on Multimodal Large Language Models in Clinical Practice, CMLLMs 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025
Country/Territory	Korea, Republic of
City	Daejeon
Period	23/09/25 → 27/09/25

Keywords

Agentic AI
Autonomy boundaries
Conversational agents
Human-centred AI
Human-computer interaction
Trustworthy AI

Access to Document

10.1007/978-3-032-06004-4_8

Cite this

Santa Cruz, B. G., Vega, C., Santangelo, P., & Satagopam, V. (2025). Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health. In J. Qiu, J. Wu, C. Langlotz, B. Huang, Z. Lei, H. Wu, H. Liu, & W. Xie (Eds.), AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings (pp. 74-84). (Lecture Notes in Computer Science; Vol. 16147 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-032-06004-4_8

Santa Cruz, Beatriz Garcia ; Vega, Carlos ; Santangelo, Philip et al. / Beyond Engagement : A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health. AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings. editor / Jianing Qiu ; Jinlin Wu ; Curtis Langlotz ; Baoru Huang ; Zhen Lei ; Honghan Wu ; Hongbin Liu ; Weidi Xie. Springer Science and Business Media Deutschland GmbH, 2025. pp. 74-84 (Lecture Notes in Computer Science).

@inproceedings{51468fc9565a468786e412c11e6a82d9,

title = "Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health",

abstract = "The rapid rise of agentic AI in mental health exposes a misalignment between conversational fluency and user safety. We present a multidimensional framework spanning nine evaluation domains—from clinical validity to relational risk and regulatory compliance—designed to assess agentic systems beyond surface metrics. Applying it to eight real-world tools reveals that the most popular agents often lack basic safeguards, while clinically validated solutions remain underutilized. These findings highlight the need for structured oversight and public awareness. Our framework offers a foundation for safer deployment, informed regulation, and design practices grounded in mental health ethics.",

keywords = "Agentic AI, Autonomy boundaries, Conversational agents, Human-centred AI, Human-computer interaction, Trustworthy AI",

author = "\{Santa Cruz\}, \{Beatriz Garcia\} and Carlos Vega and Philip Santangelo and Venkata Satagopam",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 1st International Workshop on Agentic AI for Medicine, Agentic AI 2025, 1st International Workshop on Clinical-Driven Robotics and Embodied AI Technology, CREATE 2025, and 1st International Workshop on Multimodal Large Language Models in Clinical Practice, CMLLMs 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025 ; Conference date: 23-09-2025 Through 27-09-2025",

year = "2025",

month = sep,

day = "22",

doi = "10.1007/978-3-032-06004-4\_8",

language = "English",

isbn = "9783032060037",

series = "Lecture Notes in Computer Science",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "74--84",

editor = "Jianing Qiu and Jinlin Wu and Curtis Langlotz and Baoru Huang and Zhen Lei and Honghan Wu and Hongbin Liu and Weidi Xie",

booktitle = "AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings",

address = "Germany",

}

Santa Cruz, BG, Vega, C, Santangelo, P & Satagopam, V 2025, Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health. in J Qiu, J Wu, C Langlotz, B Huang, Z Lei, H Wu, H Liu & W Xie (eds), AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings. Lecture Notes in Computer Science, vol. 16147 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 74-84, 1st International Workshop on Agentic AI for Medicine, Agentic AI 2025, 1st International Workshop on Clinical-Driven Robotics and Embodied AI Technology, CREATE 2025, and 1st International Workshop on Multimodal Large Language Models in Clinical Practice, CMLLMs 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025, Daejeon, Korea, Republic of, 23/09/25. https://doi.org/10.1007/978-3-032-06004-4_8

Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health. / Santa Cruz, Beatriz Garcia; Vega, Carlos; Santangelo, Philip et al.
AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings. ed. / Jianing Qiu; Jinlin Wu; Curtis Langlotz; Baoru Huang; Zhen Lei; Honghan Wu; Hongbin Liu; Weidi Xie. Springer Science and Business Media Deutschland GmbH, 2025. p. 74-84 (Lecture Notes in Computer Science; Vol. 16147 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Beyond Engagement

T2 - 1st International Workshop on Agentic AI for Medicine, Agentic AI 2025, 1st International Workshop on Clinical-Driven Robotics and Embodied AI Technology, CREATE 2025, and 1st International Workshop on Multimodal Large Language Models in Clinical Practice, CMLLMs 2025, Held in Conjunction with the 28th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2025

AU - Santa Cruz, Beatriz Garcia

AU - Vega, Carlos

AU - Santangelo, Philip

AU - Satagopam, Venkata

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

PY - 2025/9/22

Y1 - 2025/9/22

N2 - The rapid rise of agentic AI in mental health exposes a misalignment between conversational fluency and user safety. We present a multidimensional framework spanning nine evaluation domains—from clinical validity to relational risk and regulatory compliance—designed to assess agentic systems beyond surface metrics. Applying it to eight real-world tools reveals that the most popular agents often lack basic safeguards, while clinically validated solutions remain underutilized. These findings highlight the need for structured oversight and public awareness. Our framework offers a foundation for safer deployment, informed regulation, and design practices grounded in mental health ethics.

AB - The rapid rise of agentic AI in mental health exposes a misalignment between conversational fluency and user safety. We present a multidimensional framework spanning nine evaluation domains—from clinical validity to relational risk and regulatory compliance—designed to assess agentic systems beyond surface metrics. Applying it to eight real-world tools reveals that the most popular agents often lack basic safeguards, while clinically validated solutions remain underutilized. These findings highlight the need for structured oversight and public awareness. Our framework offers a foundation for safer deployment, informed regulation, and design practices grounded in mental health ethics.

KW - Agentic AI

KW - Autonomy boundaries

KW - Conversational agents

KW - Human-centred AI

KW - Human-computer interaction

KW - Trustworthy AI

UR - https://www.scopus.com/pages/publications/105018303446

U2 - 10.1007/978-3-032-06004-4_8

DO - 10.1007/978-3-032-06004-4_8

M3 - Conference contribution

AN - SCOPUS:105018303446

SN - 9783032060037

T3 - Lecture Notes in Computer Science

SP - 74

EP - 84

BT - AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings

A2 - Qiu, Jianing

A2 - Wu, Jinlin

A2 - Langlotz, Curtis

A2 - Huang, Baoru

A2 - Lei, Zhen

A2 - Wu, Honghan

A2 - Liu, Hongbin

A2 - Xie, Weidi

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 23 September 2025 through 27 September 2025

ER -

Santa Cruz BG, Vega C, Santangelo P, Satagopam V. Beyond Engagement: A Multidimensional Framework to Evaluate the Safe Development of Agentic AI in Mental Health. In Qiu J, Wu J, Langlotz C, Huang B, Lei Z, Wu H, Liu H, Xie W, editors, AI for Clinical Applications - 1st International Workshops, Agentic AI 2025, CREATE 2025, and Clinical MLLMs 2025, Held in Conjunction with MICCAI 2025, Proceedings. Springer Science and Business Media Deutschland GmbH. 2025. p. 74-84. (Lecture Notes in Computer Science). doi: 10.1007/978-3-032-06004-4_8