Large Language Models in the Field of Suicide Prevention

What Suicide Prevention Research Can Teach Us About the Future of AI-Assisted Threat Assessment

New research highlights both the promise and the responsibility of using AI in prevention-focused work.

A recent scoping review published in the Journal of Medical Internet Research examined how large language models (LLMs) are being used in suicide prevention and self-harm research. The review analyzed 43 studies and identified a rapidly growing body of work exploring how AI can support earlier identification of risk, improve access to support, and assist professionals working in prevention-focused environments.

While the study focused on suicide prevention rather than behavioral threat assessment, many of the findings are directly relevant to those working in workplace violence prevention, school safety, threat management, and public safety.

AI Is Becoming a Prevention Tool

One of the clearest findings from the review is that AI may help organizations process large amounts of information more efficiently than traditional approaches alone. Researchers found growing interest in using LLMs to identify concerning language patterns, support triage processes, enhance training, and improve access to resources.

For professionals working in Behavioral Threat Assessment and Management (BTAM), this raises an important question:

What if AI could help practitioners identify emerging patterns earlier, organize information more effectively, and support more consistent documentation without replacing human judgment?

That question sits at the heart of many current efforts to responsibly apply AI in violence prevention.

The Real Opportunity Is Augmentation, Not Automation

The most important takeaway from the review is not that AI can replace professionals.

It cannot.

Researchers repeatedly emphasized the need for multidisciplinary collaboration, high-quality data, human oversight, and careful evaluation of model limitations and biases.

This mirrors what experienced threat assessment professionals already know. Context matters. Human behavior is complex. Decisions involving safety, intervention, and care cannot be delegated to an algorithm.

The strongest role for AI is as a decision-support tool.

AI can help organize information, identify patterns, summarize large volumes of data, surface questions that deserve attention, and support documentation. Human practitioners remain responsible for interpretation, assessment, and intervention planning.

Why This Matters for BTAM

Behavioral threat assessment teams often face challenges that AI is uniquely positioned to help with:

• Reviewing large volumes of written information

• Identifying recurring themes and behavioral patterns

• Tracking changes over time

• Organizing fragmented information from multiple sources

• Supporting consistent case documentation

• Highlighting information gaps that require follow-up

These tasks consume significant time and resources. AI may help teams spend less time sorting information and more time evaluating it.

Importantly, this does not mean predicting violence. It means supporting prevention.

The Ethical Questions Matter

The review also highlights concerns that should not be ignored.

Researchers identified privacy, security, transparency, bias, and accountability as critical issues that must be addressed before AI can be responsibly deployed in sensitive prevention settings.

These concerns are especially relevant in BTAM.

Organizations must understand:

• What information AI is using

• How recommendations are generated

• What limitations exist

• Where human review is required

• How data is stored and protected

Any prevention-focused AI system should be designed around transparency, explainability, and human oversight.

Looking Ahead

The research community is increasingly exploring how AI can support the prevention of self-harm, suicide, family violence, and other forms of interpersonal violence.

For the BTAM field, the opportunity is not to replace structured professional judgment.

The opportunity is to strengthen it.

Used responsibly, AI may help practitioners recognize patterns earlier, improve consistency, reduce administrative burden, and focus more attention on the people and situations that need it most.

The future of prevention is unlikely to be human or artificial intelligence alone.

It will be human expertise supported by carefully designed AI tools that help professionals make better-informed decisions while keeping people, ethics, and safety at the center of the process.