Your use of AI

This page explains how Conducttr’s AI features should be used. It outlines what the AI is for, its limitations, the risks of misuse, and the oversight expected from customers. All users should review this guidance before using AI within Conducttr.

Intended Purpose and Known Limitations

Intended Purpose

The AI features in Conducttr are designed exclusively to support exercising, training, rehearsal, and simulation conducted within the Conducttr platform. Permitted uses include:

• Generation of simulated exercise content such as serials, injects, articles, imagery, audio, and scripted communications

• Generation of qualitative assessments of player-produced content for training and evaluation purposes (which should be reviewed by a qualified observer)

• Creation of realistic, and where required, confronting or sensitive material within a controlled exercise environment

• Generation of cognitive models for personas

The AI features are not designed for operational decision-making, real-world publication, live public communications, or external commercial exploitation.

Boundaries

• AI-generated content may be inaccurate, incomplete, misleading, biased, offensive, disturbing, or unlawful

• Outputs are probabilistic and do not constitute verified fact, expert advice, or legal guidance

• No content moderation, filtering, or ethical guardrails are applied

• AI outputs are provided on an “as is” basis with no warranties as to quality, suitability, or fitness for purpose

• AI systems may reflect limitations or biases inherent in training data or prompt design

All outputs must be treated as unverified draft material requiring human review.

Reasonably Foreseeable Misuse and Risk Mitigations

Our Service Agreement prohibits the following foreseeable misuses

• Publishing AI-generated content outside the exercise environment

• Using AI outputs to inform real-world operational, political, financial, legal, or safety decisions

• Generating content for harmful, deceptive, unlawful, or unethical purposes

• Failing to review outputs before participant exposure

• Treating AI assessments as authoritative conclusions

Risk Mitigations

Customers must implement appropriate governance and oversight measures, including:

• Mandatory human review and approval before content release within exercises

• Clear labelling of AI-generated material within internal workflows

• Separation between exercise systems and live operational systems

• Access controls restricting AI use to authorised personnel

• Internal policies defining acceptable use and escalation procedures

• Audit logging of prompts and outputs where appropriate

Responsibility for lawful and ethical use rests entirely with the Customer.

Required User Configurations and Recommended Human Oversight

Required Configurations

• AI must only be used within Conducttr-hosted exercise environments

Recommended Human Oversight Measures

• Designate a responsible Exercise Director or AI Content Reviewer

• Implement a two-person review process for sensitive or high-impact outputs

• Conduct pre-exercise validation testing of content

Bring Your Own Prompt – Customer Responsibilities

With our Bring Your Own Prompt feature, the Customer supplies a custom prompt to be used by the AI to review and assess content produced by the training audience. The Customer is solely responsible for the design, wording, and governance of that prompt.

The Customer must ensure that:

• The prompt is clearly aligned to the exercise objectives and assessment criteria

• The prompt does not instruct the AI to make unsupported factual claims, legal conclusions, or definitive judgments

• The prompt avoids biased, discriminatory, or inappropriate framing

• All AI-generated assessments are reviewed by a qualified human before feedback is issued or relied upon

Best Practice

• Define the evaluation criteria explicitly within the prompt

• Keep assessment categories structured and measurable where possible

• Test the prompt on sample audience content before live deployment

• Apply a moderation or quality check process before sharing AI-generated evaluations with participants

• Maintain version control and restrict editing rights to authorised exercise staff

TSL does not review or validate Customer-supplied prompts. Responsibility for the framing of assessments and any resulting feedback rests entirely with the Customer.