What Broke and What Could Be Built: A Technologist’s View on 345 Park Avenue

On July 28, 2025, New York City was faced with a tragedy that resulted in the loss of lives. A gunman walked into 345 Park Avenue carrying a rifle. Four people never made it home. Among them were a lobby security professional and an off-duty NYPD officer who ran toward danger.

Reflecting on the role of technology in the context of a tragedy like this is complex and necessarily incomplete. Still, as a technologist leading a company whose mission is precisely to help organizations avoid incidents like these, it's my responsibility to reflect on the role that a modern security stack can play to detect earlier, notify faster, and constrain movement more effectively, which could help in future incidents. We live in a privileged time in which technological advancements in physical AI, cloud connectivity, edge computing, and robotics can be effectively integrated to serve as a force multiplier for security teams in charge of the safety of employees and bystanders.

This article draws on information that is publicly reported and on patterns we see across urban high-rises. My goal is to provide practical, layered improvements that can be adopted to reduce, if not eliminate, such occurrences.

To be clear, 345 Park Avenue was not an unsecured facility. It had multiple layers of protection in place, consistent with what you’d expect from a high-profile Midtown Manhattan tower housing tenants like Blackstone and the NFL:

• Surveillance cameras monitoring entry points and public spaces
• Armed, off-duty NYPD officer stationed in the lobby
• Glass turnstiles requiring badge access to enter elevator banks
• Modern access control system for floors and tenant suites
• Bathrooms that doubled as safe rooms with bulletproof doors

These components reflect a layered and modernized security environment, designed to enforce access control, detect threats, and support situational awareness. And yet, on July 28, those systems were circumvented.

The tragedy highlights a difficult truth: even well-secured buildings can be vulnerable if security systems are responsive, rather than proactive. Video surveillance, access controls, and the people in security teams must be supported by accurate, real-time intelligence and automation to match the speed and unpredictability of modern threats.

1. Programmatic, Early Threat Detection

The attacker arrived by car, parked, stepped onto the curb of Park Avenue with a rifle in his hand, and moved towards the building’s glass revolving door and lobby. The first 911 calls came at around 6:28 PM.

Exterior cameras and sensors already monitor curb lanes, plazas, and building perimeters, as evidenced in reports following the incident, which show the previous image of the gunman carrying the rifle. And while the shooter was in fact flagged on video with threat-detection software, this occurred at around 6:27 PM, approximately one minute before police were first notified. This means that early warning did not translate into rapid lockdown or intervention. The detection system failed to trigger timely human response protocols. Why? What we usually see with other customers is that, in a city like New York, defined by constant motion and dense foot traffic, traditional threat detection systems are overwhelmed by the sheer volume of activity, creating too many alerts to monitor effectively. Security teams are overflooded by the relentless background noise of urban life.

Today, it is possible to upgrade existing security camera systems with AI-powered software capable of intelligent, context-aware threat detection, operating with systematic precision and unwavering vigilance without overhauling the existing infrastructure. This enables security teams to receive alerts for genuine threats, in real time, cutting through the noise of routine, non-threatening activity that would otherwise flood their attention. In a scenario like this, such a system could have flagged the unmistakable visual signature of a rifle the moment it appeared on camera. That alert could have reached the guard desk and Security Operations Center (SOC) within seconds, providing critical lead time to initiate security protocols, and potentially, provide a real opportunity for prevention.

2. Automated, Adaptive Access Controls

While 345 Park Avenue had modern access control infrastructure in place (glass turnstiles, badge-based elevator access, and staffed security posts), those systems were ultimately static in nature. They were designed to regulate routine movement, not dynamically respond to unfolding threats. Once the attacker bypassed the turnstiles and entered the elevator, there were no systems in place to remotely restrict vertical movement or lock down individual floors in real time.

But what if more responsiveness had been built in? Imagine that the security team had used those 60 seconds of forewarning to activate an Active Threat protocol, triggering an immediate building-wide lockdown. Today, advanced door lockdown technologies are deployed in commercial, educational, and high-security environments. Solutions like Kisi, Brivo, and ICT Protege enable security teams to remotely and instantly lock doors using cloud-based platforms and mobile apps. Through secure hardware retrofits or smart locks, they give teams the ability to initiate site-wide lockdowns within seconds, without requiring physical presence.

Beyond site-wide lockdowns, modern access control is evolving toward intelligent orchestration and adaptive response. Platforms like Genetec Security Center and LenelS2 now offer zone-specific lockdown capabilities, enabling security teams to isolate threats to a particular floor, wing, or sector, while keeping safe areas accessible for evacuation or first responders. This level of granularity transforms lockdowns from blunt-force measures into targeted containment strategies. In this incident, the shooter was able to access the elevator and reach the 33rd floor, where the attack continued. Had the security team been equipped with the ability to remotely restrict elevator access and secure upper floors, they could have contained the attacker’s movement to the lobby, shielding the majority of building occupants. Additionally, when you cut the attacker’s floor options, you shrink the search grid for arriving officers, and precious time is saved in the overall response timeline.

3. Building-wide, Multi-tenant Communication

One often-overlooked vulnerability in high-rise, multi-tenant buildings, such as 345 Park Avenue, can be the absence of a centralized communication channel. Unlike single-employer campuses, there is no shared Slack workspace, email group, or emergency protocol that spans across floors and tenants. During an unfolding threat, individuals are left to make life-or-death decisions, whether to shelter in place or flee, based on unreliable cues.

This gap in communication can be bridged with the deployment of a mass notification system. Platforms like Alertus, Everbridge, and Singlewire InformaCast can issue coordinated messages via overhead speakers, SMS, email, and collaboration tools like Microsoft Teams or Slack, all within seconds of a verified threat.

Imagine if, in this incident, a pre-programmed alert had been triggered within 10 seconds of the firearm being detected. Clear, intelligible voice announcements in lobbies and elevator bays could have deterred movement toward danger, while simultaneous digital notifications could have guided occupants to safer zones. People hesitating at a stairwell, hovering near an elevator, or deciding whether to exit their office would have had immediate, reliable information.

In addition, integration with real-time location systems (RTLS), such as those from Zebra Technologies or GuardRFID, can enhance situational awareness by tracking staff and visitor locations throughout the building. These systems allow security teams to understand who is in harm’s way and help guide emergency responders to high-risk areas even more efficiently.

4. Balancing Intelligence, Safety, and Privacy

While the promise of real-time AI-enabled response becomes a reality, several guiding principles can help shape thoughtful implementation, especially in sensitive, high-traffic environments like NYC.

The first consideration is the role of human judgment. In any security model, AI should act as a superhuman assistant. Even with high-confidence alerts, human verification remains essential. Models with thresholds for confidence levels, escalation paths, and approval checkpoints before triggering automated actions are necessary.

Another area of focus is life-safety. Lockdown sequences, for example, must account for fire service recall, free egress, stairwell re-entry, and first responder access. All factors that are often governed by code and local ordinance.

Finally, any deployment of advanced technology must implement privacy by design. While AI-powered video analysis can detect threats with speed and accuracy, we believe it’s necessary to do so without relying on facial recognition. Respecting the boundaries of individual privacy is foundational to building trust with the people to protect.

A Possible Path Forward

We can’t know with certainty how the outcome on July 28 might have changed, even if these technologies had been in place. But that uncertainty shouldn’t deter us from pursuing smart, integrated approaches to physical security. Ones that enhance safety without compromising personal privacy or autonomy.

The ideas outlined here are not prescriptions, but working hypotheses, meant to provoke thought, inspire collaboration, and advance the conversation around how organizations can respond when every second matters. In an era of rising violence, even in the workplace, the responsibility to adapt is shared.

The goal is not to harden our cities into fortress-like structures, but to enable the systems we already rely on—cameras, access control, communications—to operate in quiet, intelligent coordination.