Rising Liability: Why Facilities Need AI Gun Detection

Facilities leaders are being asked to do the impossible: keep people safe in open, high-traffic environments, without turning the workplace, hospital, campus or venue into a fortress. Cameras and guards help, access control helps and policies help. But the liability question that keeps popping up after a major incident is simpler and tougher:

Could you have detected the threat sooner and acted faster before the first shots were fired?

It’s not just public scrutiny anymore. Lawmakers are starting to formalize earlier detection expectations, especially in high-risk environments, making “time-to-detection” a growing compliance and liability issue, not just a best practice.

When Detection Comes Too Late

That “time-to-detection” problem is exactly what brought renewed attention to 345 Park Avenue in Midtown Manhattan, where a gunman approached the building with an assault-style rifle, entered the lobby, began shooting, then used the elevator to get to the 33rd floor and continued the attack before taking his own life. 

In December 2025, the incident turned into a bigger discussion about expectations for security programs when a wrongful-death lawsuit was filed in the New York State Supreme Court. Reports say the complaint alleges that no security measure deterred, detected or delayed the attacker despite the weapon being visible, and points to gaps in weapon detection, monitoring/interception protocols, communication systems, training and coordination. The reporting also highlights a critical operational detail: the building had an elevator-freeze function, but the officer responsible for activating it was shot before he could do so. 

For security and facilities teams, the takeaway isn’t “this could happen anywhere” (you already know that). The Propmodo article argues that security is increasingly seen as a system, “from public plazas to interior circulation like elevators.”

Data Supports a Move to Earlier Detection

Violence risk isn’t a trend line. It’s a measurable impact on workplaces and communities year after year:

• 42,886 people died from gun-related injuries in the U.S. in 2024 
• The U.S. Bureau of Labor Statistics reports 57,610 nonfatal workplace violence cases (private industry) in 2021–2022 severe enough to require days away from work, job restriction or transfer (DART). 
• In 2022, BLS counted 524 workplace homicide fatalities, and 83% of them were due to gunshot wounds. 
• The FBI had 24 active shooter incidents in 2024. 

Those numbers don’t mean every facility needs the same security posture. They do mean one thing: the most valuable seconds are often the ones before an incident escalates when a weapon becomes visible, but harm may still be preventable or reducible with immediate action. Early detection through AI gun detection enables a rapid response, helping to reduce harm.

Why Facilities Struggle with Time-to-Detection

Most organizations already have cameras. The problem is that cameras are often treated as forensic tools, useful only after the fact, because continuous monitoring at scale is challenging. Even strong security programs can face real-world constraints:

• High-traffic lobbies, plazas and entrances create momentary blind spots
• Guards and reception teams can be overwhelmed by normal volume
• Threats move fast through transition points (doors, turnstiles, elevators, stairwells)
• “Push-button” procedures (like elevator recall/freeze) rely on a person being able to act under extreme stress, something the 345 Park reporting suggests may fail in the worst moments 

This is where AI gun detection fits, not as a replacement for people or plans, but as a way to detect a visible firearm quickly enough to trigger response workflows. 

From Visible Weapon to Verified Response in Seconds

AI gun detection uses computer vision to analyze video feeds for visible firearms and flags likely threats in real time. The goal is to compress the time between a weapon becoming visible, the threat being recognized and a response being initiated. 

Omnilert’s AI Gun Detection System is designed around that timeline. Our AI identifies a potential gun threat in fractions of a second and automatically initiates a security response workflow that includes real-time human verification and alerts first responders and safety systems through integrations. The system features advanced firearm detection capabilities, including identifying guns in different lighting conditions and across a wide detection range. 

That “human verification” piece matters. In practice, it helps teams act quickly without turning every alert into a full-scale disruption, reducing the risk of alarm fatigue and improving confidence in escalation decisions. 

Omnilert AI Gun Detection has received full DHS SAFETY Act Designation from the U.S. Department of Homeland Security (DHS) and is on the Approved Technologies List, which can strengthen defensibility by providing liability protections for certain claims connected to an “Act of Terrorism,” as determined by DHS.

A realistic expectation (and an important one for stakeholder trust): visual AI works best when a weapon is visible to a camera. Good deployments start by ensuring coverage of the spaces where threats are most likely to appear and move quickly, including entries, lobbies, perimeter approaches and vertical circulation points.

The First 30 Seconds: A Scenario That Shows the Difference

Imagine a multi-tenant office building during peak arrival time.

A person enters from the plaza with a visible weapon. Instead of relying on someone at the desk to notice it in the moment, or hoping a guard catches it while juggling multiple tasks, AI gun detection flags the threat on camera and triggers a workflow:

Security receives an alert with context, the incident is quickly verified, and leadership and staff receive clear, immediate instructions. Security can implement pre-planned actions (like restricting access pathways, initiating targeted notifications and coordinating with law enforcement), while occupants move away from the threat and shelter appropriately.

Not every incident can be prevented. But compressing recognition and coordination into seconds can change outcomes, especially in those “plaza to lobby to elevator” moments highlighted in the 345 Park reporting. 

Duty of Care, Continuity and Rising Expectations

Facilities don’t buy safety technology because it’s interesting. They buy it because it supports a safer environment and reduces operational and organizational risk.

The 345 Park lawsuit coverage puts a spotlight on a shift many teams already feel: scrutiny (and potential liability) is moving from “did you have something?” to “did your measures work together quickly enough?” In practice, that means speed and integration matter – how quickly you can detect, verify, communicate, and act.

That shift isn’t happening in a vacuum. Regulators are beginning to translate “reasonable protection” into concrete requirements, especially in settings where the risk profile is well established.

A clear example is California’s AB 2975, passed in 2024, which requires Cal/OSHA to adopt standards (by March 1, 2027) mandating that hospitals implement a weapons detection screening policy. The policy must include automatic weapons detection devices (not handheld wands) at specified entrances (including the main public entrance, emergency department entrance, and labor & delivery when separate), trained personnel to operate the devices, public signage, and defined response procedures. While AB 2975 is healthcare-specific, it is a sign of a broader trend: weapon detection and response workflows are moving from optional upgrades to expectations in high-risk environments.

The implication for facilities leaders goes beyond healthcare: expectations are moving toward earlier detection and integrated response as the standard of care. As more jurisdictions and industries codify these requirements, the question of liability becomes less about whether you intended to keep people safe and more about whether your security system was designed to detect fast enough to trigger action.

That shift is why more organizations are evaluating camera-based weapons detection as part of a connected security strategy. The argument is straightforward: if a firearm is visible on approach or at an entry point, relying solely on human observation creates a fragile dependency, especially during peak traffic or when staff are managing multiple tasks at once. Technology that can detect a visible weapon quickly, route the alert for real-time verification, and immediately trigger notifications and response actions helps reduce that dependency and improve speed and coordination. 

The practical risk for organizations isn’t only the incident itself. It’s the post-incident question of whether your security program was reasonable given the environment, the known threat profile, and the tools available. As expectations shift toward faster detection and integrated response, “we had cameras” is less persuasive if those cameras weren’t paired with a way to recognize a visible threat and initiate action quickly.

From a facilities perspective, the case for technology like Omnilert’s typically comes down to three outcomes:

1. Safety outcomes: earlier awareness can support faster response and clearer instructions when it matters most. 
2.  Operational resilience: a faster, more targeted response can reduce confusion, downtime and disruption during and after an incident.
3. Risk posture: showing proactive measures and integrated response workflows strengthens duty-of-care planning and improves defensibility after an incident.

How To Get Started: A Practical Deployment Approach

If you’re considering AI gun detection for corporate campuses, healthcare, higher ed, venues or multi-tenant properties, a good deployment usually looks like:

• Map the highest-value camera views: entrances, lobbies, plazas/perimeters and vertical circulation points (elevators/stairs). 
• Define your first 30-seconds playbook: who gets notified, who verifies, what actions are triggered.
• Connect detection to action: alerting to your communication and security workflows so you’re not stitching together tools mid-incident. 
• Train and drill: technology works best when people know exactly what to do when an alert arrives.
• Maintain readiness: test workflows, keep software updated, and drill teams regularly

Bottom line

The 345 Park Avenue lawsuit coverage is a stark reminder that threats can come from anywhere and arise in seconds, and that even good intentions can fail when everything depends on one person being able to act under pressure. 

Facilities don’t need more fear. They need time and a way to turn time into action. And increasingly, they need documentation and defensibility, because expectations are shifting from “having cameras” to having a connected security system that can detect, verify, and initiate response fast enough to matter. That’s where regulations like California’s AB 2975 signal the direction the market is heading: earlier detection is being treated as an operational requirement, not an optional upgrade.

That’s the core case for AI gun detection like Omnilert’s: detect a visible firearm fast, verify in real time and initiate a response workflow fast enough to matter. 

If you’re rethinking how quickly your facility can detect and respond to a visible weapon, Omnilert can help. Contact our team to see how AI Gun Detection with real-time human verification works with your existing cameras, and to discuss a deployment plan for your facility.

Frequently Asked Questions FAQs