Ai Ethics of Ambulance Bay Screening Technology

What is the purpose of the AI in your application?
Athena uses AI-enabled computer vision and detection technology to help identify potential concealed contraband during entry screening. The purpose is to improve situational awareness, support safer and more consistent security operations, and help security personnel focus attention on items or events that may require secondary screening. Athena’s AI is not used for diagnosis, treatment, clinical triage, or other patient-care decisions.

What decisions or processes does your AI support?
The AI supports the security-screening process by identifying potential anomalies or threat indicators for human review. It may help prioritize secondary screening, guide officers to the location of a potential item, and create an operational record of screening activity and response. It does not independently decide whether a person may enter, must be detained, or poses a threat.

Does your AI augment human work or replace decision-making?
Athena’s AI augments human security personnel and does not replace it. The design to assist a trained operators by surfacing potential areas of concern; final assessment, secondary screening, escalation, access-control, and safety-response decisions remain with authorized human personnel and the customer’s policies.

Is human review required before AI outputs are used?
Human review is required. The AI-generated alerts are intended to be reviewed by trained personnel before operational action is taken. Athena does not authorize automated enforcement, detention, denial of access, clinical action, or other consequential action solely on the basis of an AI alert.

At what point can a user validate, edit, or override AI output?
Users can validate or override the AI output at the point of alert review and secondary screening. A trained officer may determine that an alert is not actionable, conduct additional screening, follow site-specific protocols, or escalate the event as appropriate. Human judgment governs the final outcome.

Are users trained to critically evaluate AI results?
Athena provides technical training and operational guidance intended to help users understand the system’s capabilities, limitations, alert-review process, secondary-screening procedures, and escalation protocols. Customers should ensure that personnel are trained not to treat alerts as conclusive and to apply judgment consistent with their security policies, legal obligations, and facility procedures.

What types of data does the AI use?
Athena’s detection technology uses imaging radar data for the computer vision A.I.

How do you ensure the data is used ethically and appropriately?
Athena’s intended use is limited to physical-security screening, safety operations, and related system performance at the device. Data use is governed by contractual commitments, customer configuration, access controls, security practices, and applicable law.

Is data used to train models, and if so, is it de-identified or consented?
Athena does NOT use end-user data to train models, unless explicitly approved. We collect training data in our Jacksonville, FL lab. Athena should only use customer or deployment data for model improvement where authorized by applicable agreements, customer instructions, and law. Where data is used for development or improvement, Athena should apply appropriate safeguards, such as de-identification, minimization, access controls, and contractual restrictions, as applicable. Customer-specific data should not be repurposed beyond authorized uses.

How do you identify and mitigate bias in your AI?
Athena imaging radar makes no attempt to understand gender, race, ethnicity or any other group information. Athena’s AI is designed to identify objects, potential weapons or security-related objects, rather than to evaluate a person’s character, intent, health status, race, ethnicity, gender, religion, or other protected characteristics. Athena mitigates bias through product testing, monitoring, human review requirements, system design focused on object and threat detection, and operational controls that prohibit consequential decisions based solely on AI output.

How do you address incorrect or misleading outputs, including hallucinations?
Athena’s detection technology is not a generative AI system that produces open-ended text or clinical recommendations; therefore, traditional “hallucinations” are generally not applicable. The relevant risk is a false positive, false negative, or otherwise incorrect alert. Athena addresses this through trained human review, secondary screening, customer operating procedures, testing, system monitoring, and product improvement processes. No alert should be treated as conclusive without human assessment.

How do you test for unfair or discriminatory outcomes?
Athena evaluates performance through testing and validation processes appropriate to its intended security use case, including monitoring detection performance and alert behavior across relevant operating conditions. Because the system is not intended to make identity-based or clinical decisions, fairness controls emphasize avoiding use of protected characteristics in decision-making and maintaining human review before consequential actions occur.

How do you ensure AI outputs are accurate and complete?
Athena uses testing, quality controls, performance monitoring, software and model updates, and trained-operator workflows to support reliable operation. However, no security-screening technology can guarantee perfect detection in every environment or circumstance. Accuracy depends on appropriate installation, calibration, operating conditions, maintenance, user training, and compliance with secondary-screening procedures.

Are AI-generated outputs clearly labeled?
Yes. Alerts and system-generated indicators are presented as AI-assisted or system-generated information for operator review. They are not presented as definitive findings, clinical conclusions, or autonomous decisions.

How do you prevent copyright issues or improper reuse of content?
Athena’s AI is not intended to generate copyrighted text, images, or other creative content for users. System outputs are operational security alerts, device data, and audit information. Athena uses contractual, technical, and governance controls to limit data use to authorized purposes and to protect customer information from improper access, disclosure, or reuse.

How are ethical concerns or issues with AI addressed over time?
Athena addresses AI ethics through ongoing product governance, customer feedback, security and privacy review, incident and issue-management processes, testing and monitoring, model and software updates, and review of evolving legal and industry expectations. Material concerns involving accuracy, safety, privacy, bias, or misuse are evaluated through appropriate technical, operational, and governance channels, with corrective actions taken when warranted.