The Trust Gap and its Causes 

Although public-facing LLMs such as ChatGPT and Perplexity are ubiquitous, they are most commonly used to address personal issues. Asking a chatbot to create a shopping list or provide feedback on an interpersonal issue requires a degree of trust, but frequent users commonly observe how often these models make mistakes in low-risk situations like these.

However, the AI agents and systems involved in security, policing and emergency response are far more complex than most members of the public can access. As these technologies are built with such unique purposes and functionalities in mind, they fall into the larger blind spot of AI illiteracy. System specs do little to sway public perception when people’s objections concern fundamental rights and safety, such as:

• Surveillance overreach and predictive policing
• A well-documented tendency for AI models to amplify the bias of creators    
• A lack of accountability when decisions are made without human input

Overall, the general public strongly supports greater AI regulation. 87% of the public favors laws that combat AI-generated misinformation and only 43% believe current legislation is adequate. 

Trust issues also become exacerbated when someone who lacks AI literacy reaches a position of power. Many respondents in the KPMG study cited above either misused AI or observed a coworker doing so, such as by uploading sensitive personal or company information to public chatbots. 

Machine learning is complex, and navigating a company’s terms and conditions can be even harder. Available findings indicate that there’s a strong need for internal and external AI education to evangelize the kind of knowledge that’s vital to building trust.

The Reality of Operations  

AI agents in security and public services have become productivity tools that enable staff to meet administrative demands and return to proactive work. Public safety and security agencies have been using them more frequently, primarily because of labor shortages and increasing workloads.

Police officers are deskbound 30–40% of the time to write reports and launch enquiries, but AI agents can reduce this time substantially by aggregating data and providing additional context that would otherwise require hours to gather manually. They’re not generating anything new; rather, this work is more closely related to the type of data analysis from which machine learning originated.

Popular applications of AI in traditional security settings include smart cameras that use adaptive algorithms to spot patterns that deviate from expected behavior. They can help:

• Observe when staff aren’t following PPE requirements
• Alert security teams of potential loitering or the presence of suspicious packages
• Reduce time spent staring at monitors

Again, we can see that these systems are built to augment human activities, not to autonomously make decisions for individuals. Building public trust in AI requires frequently reiterating that humans remain firmly in the loop at all times. 

Visible Safeguards Are About more than Optics

Security systems with AI agents can reach the point where they are safer on average than those with humans alone, but it will not matter to the general public if transparency is not built in at a structural level. The observation of national and international data protection laws provides a baseline of trustworthiness, but as more people call for stricter regulation of the technology, it becomes increasingly necessary to exceed public expectations. 

Some applications of AI, such as those seen in policing and security, draw information from multiple sources, helping to mitigate fear of bias. Beyond this, such systems must take strong and open stances on: 

• Strong data retention and minimal collection policies
• Clearly defined use cases and user controls
• Accessible audit trails and access logs 

To aid the building of public trust surrounding the implementation of AI technology, the public’s primary concerns must be addressed. Currently, many people share sensitive information with public chatbots without understanding how that data is stored or used.

Building trust requires a different path. By maintaining human-led oversight and strict data governance, our AI serves as a transparent and accountable partner in public safety. We aren't just following the conversation on AI ethics; we are setting the standard for how it is applied to protect our world.