This is the second SigInt, and I'll admit it's turning into my favorite thing to write.

Every other week I'm chasing whatever just happened, and once a month I get to put the new headlines down and go back through the ones I already ran to figure out what was going on underneath them. Some months the thread is obvious. This month it wasn't, and what I ended up with argues against the thing almost every outlet said in July.

It's at the bottom of the issue, so check it out and let me know what you think!

Every security director has a binder somewhere with a guard vendor's paperwork in it. State license, certificate of insurance, general liability and workers comp limits, maybe a bonding letter and a note that the vendor is a national name with a recognizable logo. Somebody pulled three bids, procurement confirmed the boxes were checked, the binder went in a drawer, and the vendor has been working the site ever since without anybody opening it again.

On July 24 a Dallas County jury put a price on a binder like that. The price was $604 million.

The verdict went against C.H. Robinson, the motor carrier Lupus Superior, and the truck driver, over a 2021 pileup on Interstate 20 in Mississippi where a Lupus Superior rig hit stopped traffic and six vehicles caught fire. Jennifer Lipe, Benjamin Brewer, and Rhoderick Coleman died trapped in their cars, and two other people were hurt badly enough to be named in the suit. C.H. Robinson is a freight broker, so it didn't own the truck, didn't employ the driver, and didn't load the trailer. All it did was pick the carrier.

So why is a security guy writing about a trucking incident?

Because of what C.H. Robinson said in its own defense. Its chief legal officer said Lupus Superior had safely delivered nearly 270 loads for their customers and held a Satisfactory rating from the Federal Motor Carrier Safety Administration at the time it was selected, and that the carrier still held that Satisfactory rating after a federal review of this exact crash. The government's scorecard said the carrier was fine before three people burned to death, and the same scorecard said the carrier was fine after the agency went back and looked at the wreck itself. The official screen never moved, and the jury assigned $604 million anyway.

Brokers used to have a structural argument that they couldn't be sued for this at all. The Federal Aviation Administration Authorization Act preempts state laws relating to a broker's prices, routes, or services, and brokers argued for years that a state negligence claim over which carrier they picked was exactly that kind of preempted state law. The circuits split on it. On May 14 the Supreme Court ended the argument unanimously in Montgomery v. Caribe Transport II, with Justice Barrett writing that a negligent hiring claim falls inside the statute's safety exception, which preserves state authority over motor vehicle safety, so the claim survives preemption and goes to a jury like any other negligence case.

Lipe is the first big trial verdict to land after that ruling, roughly ten weeks later, and it landed at $604 million. Anyone who read Montgomery as a narrow procedural cleanup got an answer quickly.

Strip out the freight and the holding is that when you select a vendor to perform work that can hurt somebody, you own the selection, and you own it independently of whatever that vendor does or fails to do on the job. The vendor's negligence belongs to the vendor, but the decision to put that particular vendor in that particular position is a separate act with separate liability attached to it.

Negligent selection (legally speaking) has been available against building owners, hospitals, property managers, and staffing firms for decades. What's new is a unanimous Supreme Court confirming that a federal preemption shield doesn't cover it, and a jury putting nine figures behind the idea in the same quarter. Plaintiff's firms read verdicts, they'll go looking for every industry sitting on the same posture, and contract security is ripe for picking.

Go back to the binder, because every item in it is the C.H. Robinson defense. It's a government or industry body certifying that the vendor is eligible to do the work, and eligibility isn't evidence that anybody evaluated whether this vendor could do your work, at your site, against your risk. C.H. Robinson had a better version of that binder than almost any security director in America is keeping, because an FMCSA rating is a real federal safety assessment rather than a licensure formality, and they had 270 loads of actual clean performance history. All of it, worthless in court.

Anybody who's run a contract guard force already knows how far the vendor on paper drifts from the vendor on the post, because managing that drift is most of the job. There's a training matrix in the proposal and there's the officer who got two hours of orientation because the account manager was covering a callout in another building, the contract promises weekly supervisory visits and those tend to cluster around renewal season, and somewhere in a supervisor's head there's a running list of the account's real problems that has never once been typed into anything:

Under a negligent selection theory that knowledge belongs to the organization whether or not it was ever recorded, and where there's no record, the plaintiff's account of it is the only one a jury hears.

Start a performance file on the vendor you already have, because the cheapest defensible thing available this quarter costs nothing but a recurring calendar entry. Log fill rate and no-shows monthly; whether the supervisory visits the contract promises actually happened, by date; and every incident where an officer's action or inaction came up, including the ones that resolved fine. At renewal, write down what you reviewed and why you renewed anyway, because a documented decision to keep a vendor with known problems is defensible if you can show what you are doing about the problems, while an undocumented renewal just looks like nobody was paying attention.

Then go one level past the certificate at selection. Ask for turnover on comparable accounts, training records on the officers who will actually work your site rather than the corporate curriculum, and for two references at sites your size and call them yourself. Write down that you asked, and write down what they said.

All of that measures the vendor against something other than your own recollection of how the account has gone. Negligence turns on a standard of care, which has to come from outside the organization being sued. A license establishes that a vendor is permitted to do the work and says nothing at all about what the work is supposed to actually look like.

Without a standard to point at, an organization doesn't get to define what reasonable looked like at its own site, and the plaintiff's expert shows up with a definition of their own. And as I wrote back in Issue 6, best practice doesn't fill that space either, because best practice is the general answer written for everybody and the question in the room is always what you did at your building.

C.H. Robinson is appealing, so none of this is settled yet. The structure, though, isn't unusual anywhere. It shows up any time an organization picks a vendor whose people can hurt somebody and picks them off a license and a certificate. Hospitals selecting staffing agencies for clinicians who work unsupervised. School districts selecting student transportation. Property managers selecting the tree service, the snow removal outfit, and whoever is servicing the elevators. Event operators selecting crowd management, which is contract security under a different job title.

The closest relative to C.H. Robinson isn't in security at all. It's the non-emergency medical transportation broker arranging a ride home for a discharged patient, who doesn't own the van, doesn't employ the driver, and does nothing but pick the carrier.

SCC Spectrum Security Almanacs

At a six in the morning shift change, sometimes neither officer is fully awake, and details get lost. The fire door propped at nine that nobody went back to check, the visitor flagged for a callback that never came, the camera that froze at noon, the incident that's open rather than closed. Any of it can evaporate in a thirty-second exchange between two tired people.

If the handoff is verbal and undocumented, there's no record the exchange happened at all, and that's where the liability sits.

I learned a golden rule in EMS. If it's not written down, it didn't happen. The corollary is that if it's written down, it definitely happened, and a written handoff puts you on the right side of both.

The move is a structured verbal handoff at relief. Before either officer leaves the post, the one going home walks the one coming on through every open item, out loud, in order, and both initial a sheet. Ninety seconds. Every relief, quiet shifts included.

Two things are worth measuring. Handoff completion tells you whether the exchange is happening at the post or getting filled in afterward in the parking lot. Items that cross more than one shift change tell you something else. An item on its third handoff is being passed rather than worked, and it stays that way until a name goes on it.

Apex AI: Join the waitlist

Once a month SigInt sets the new headlines aside and goes back through the Signals notes you already read, to name what they had in common that none of them showed on its own. Here's what ran across July.

A stolen trailer, a hospital floor, and a restaurant alarm log don't obviously belong in the same paragraph. What they share is that in each one the control that would have worked was already ordinary. Not new, not expensive, not something anybody had to invent. It just wasn't there.

The freight yard had a guard watching for somebody climbing a fence. What stops a fictitious pickup is a phone call to the broker before the load moves, which has been standard practice for years, and it's the easiest thing in the world to skip on a holiday weekend when the supervisor is off and the phone goes to voicemail. Los Robles had panic buttons, added after a death the year before, that nurses say they were never properly trained to use, and an alert that went out without saying that shots had been fired. Training people on a button you just installed is not an advanced practice. The restaurants had hardware built for the robbery that turns out to be 3 percent of the job, while two thirds of the real load is a guy loitering at the entrance during the dinner rush and an argument somebody has to watch long enough to judge.

What changes across the three is only how the thing went missing. In the yard it was skipped, a known step dropped when the staffing got thin. At the hospital it was installed halfway, equipment in place and the training that makes it work never following it. In the restaurants it was never treated as the main event at all, an operator's judgment carrying two thirds of the incidents while the model stays built around the rare one.

None of the three is a technology gap. The camera, the panic button, and the fence all worked. Every one of these failures happens in the space between a system reporting something and a person deciding what it means, and that space usually has nothing written about it, which is how it goes missing quietly and stays missing.

So take your top three incident types from last quarter and name the control that's supposed to catch each one. Then find out whether it exists in writing or only in somebody's habit. The ones that live in a habit are the ones that disappear on a holiday weekend, or the week the person who always did it is out.

SCC Spectrum Security Almanacs

Tuesday overnight, all five sites, nothing burning. Drawer at the grocery come up eleven short and before anybody starts, that's Donna's math, that ain't a theft. Rain got in the loading dock over at Chestnut again, third time this month, ticket's in. That's four tickets now on that one door frame if anybody's counting, which they ain't.

So Mikey's sheet. You know the one, where at relief you gotta stand there and go through every open item out loud with the guy coming on and then youse both initial it. Ninety seconds he says. I told him I been doing this forty years, I know how to tell a man what happened on my shift. He didn't care.

Anyway. Two weeks back day shift left a fire door propped open with a milk crate, somebody running smokes out the back. I flagged it, wrote it on the sheet, Ronnie initialed, ticket went in. Maintenance keeps pulling the crate and it keeps comin' back.

Last night somebody come in that same door and helped himself to about four hundred bucks of product out the back room. Crate was right where it always is.

This morning Mikey comes around asking who knew about that door. Who knew. I handed him the sheet. Two weeks of it, dated, initialed, ticket number, my handwriting. He got real quiet and went and found somebody else to talk to.

He might be surprised to know I know what I'm doin' here.

The handoff that protected Ricky was the one somebody wrote down. Make it a written standard and it protects every officer on every shift, not just the one who happened to be thorough.

SCC Spectrum Security Almanacs. www.strategiccommandcenters.com/products