AI, GSOC and Physical Security AI

Internal/external security models??

I'm still getting fully up to speed on AI - using ChatGPT for stories and trivia questions fun, but I'm grappling with how this moves to critical functions like the Global Security Operations Center. Appreciate any ideas and pointers anyone has, the below is a stream of thought (not made with ChatGPT), versus a well thought out set of ideas:

I'd like to see a Security Language Model emerge that would service the GSOC, but be interoperable with other corporate functions that have overlapping issues (e.g. dual use intelligent camera that can detect a weapon and also alert to a slip and fall injury). Seems like it would have to include:

1. Internal data - There is a plethora of data that needs to be rationalized from a variety of sources in the security realm, some that are historical and some that are real time and in continual flux. This all needs to be handled and protected, not shared out into the public LLMs. The scale of the data seems far more modest than a general purpose model.

2. External, near real time data: TX360 provides proactive identification of physical security threats from Open and other intelligence sources to corporations that want to have as much time to prepare/prevent/respond as possible (minutes in a nearby active shooter or protest, days for a slow moving hurricane). AIs that interlink on this capture capability will also become more multi-modal and capture intelligence from many sources that can be made better through data interoperability and standardization. Add in the fact that this can be done in hundreds of languages at the same time,and you can see a corporate intelligence center that has more capability than a federal agency did 20 years ago for real time analysis.

3. Crisis response capability: As you build the dynamic data sets and they are incorporated into the real time status and operation of the GSOC, better preparedness and likely a deeper set of correlated impacts could result from an AI versus humans trying to remember that event 3 years ago or scrambling for the 3 ring binders up on the top shelf. AIs will able to go deeper into impacts, well past primary, secondary, and tertiary. For example, an AI may see that fuel will be a problem for the organizations needs, and move it up the priority list days before there isn't any, allowing for hedging, external supply acquisition, etc. As the capabilities mature, the AI could query the after action teams with questions regarding future scenarios and update plans in advance.

4. Notification: Communicating critical information to the right people at the right times and in a format that can be used by them is critical, and goes far beyond the "mass" approach used today. AI will allow these messages to be delivered in multiple ways that we can't utilize today - text to video not a thing in most GSOCS. When notification/communication has to span everyone from the board of directors to the people on the front line hurricane response, it needs to be near real time, tactical, and strategic in different measures

5. Decisions: AI in the GSOC is likely to remain a partner to the humans for a significant amount of time (how long depends on the exponential speed of AI as a whole). New versions of AIs will include the ability to command a general purpose robot, so automatically deploying a drone or robot in the event of certain situations seems rational, letting an AI make life or death decisions in the event of an active shooter not so much :-) AIs recommending decisions can be very helpful if they are succinct - many of today's security AIs babble incessantly with way too many intelligence items - so adjusting recommendations to a human's capacity to comprehend and act will be an evolving challenge.

6. Interoperability: If the AIs can ultimately blend the silos inside an organization together into a very smart, cohesive whole, the entire organization will likely benefit and some more efficiencies can be realized and overall effectiveness should rise steeply. Impact on the number of jobs? No idea yet, but if the AI can capture the data, separating the great from the adequate or dismal should be much more easily accomplished. All these functions can grow into every day use; I love the concept that Microsoft has with copilot capabilities that may all be utilized with a common capability and interface. Imagine if a board of directors or C-Suite could use an interactive ChatGPT type interface to query the status of a critical event without sidelining the most important incident commanders for yet another briefing....

As mentioned, stream of thought....

Pete ODell is the CEO of www.swanislandnetworks.com. TX360 is the Physical Security Proactive Intelligence platform the company offers. Its used by corporations with assets around the world who are concerned about lives, peoperty, brand and operational continuity.