Invest1 distinct publisher3 min readUpdated
AI cameras and a multi-agency task force caught three men stripping CSX containers. The loss economics behind that response matter more than the arrests.
The Investor · Invest desk

Compiled by The InvestorSomething wrong?How this is made
AI cameras and a multi-agency task force caught three men stripping CSX containers. The loss economics behind that response matter more than the arrests.
Follow any of these and your For You feed starts watching them — no settings page required.
Three men climbed onto CSX railcars parked outside a Memphis depot in November, broke the container locks and took more than $20,000 of goods, including women's clothing bound for a Belk store, Lazer Blocks toys and two cases of Fre non-alcoholic chardonnay [1]. What separated it from dozens of similar thefts in recent years is that AI-powered cameras were watching, and a task force led by the Tennessee Bureau of Investigation, drawing on federal, state and local agencies, the Tennessee National Guard and CSX's own police force, put a plane in the air that tracked the getaway vehicle for 11 miles [2][3].
Officers moved in hours later as one of the men unloaded boxes, and two others were arrested over the following days, according to police records reviewed by Bloomberg News [4]. The thieves had hidden part of the haul in nearby woods to collect later [5].
The arrests are the least interesting part. An industry group representing CSX, BNSF and Union Pacific counts more than $200 million of goods taken in 75,000 thefts on US freight rail networks in 2025 [6]. That works out to roughly $2,700 per incident and about 205 incidents a day [7][8]. This is an attritional loss pattern, not a catastrophe one: high frequency, low severity, and therefore something that shows up in pricing and retentions rather than in headlines. Carriers are generally liable for lost goods up to a certain dollar amount, with the customer typically buying insurance above that [9]. The Memphis haul, at over $20,000, was roughly seven times the average incident [10].
The tail is fatter than the average suggests. A BNSF train lost about 1,985 pairs of unreleased Nikes worth more than $440,000 in January of last year, according to a filing in US District Court in Phoenix, or about $222 a pair [11][12]. Los Angeles police found $1.5 million of stolen cargo, including Milwaukee Tool products, in a single raid [13]. A CSX spokesperson said more than $900,000 of tires alone have gone missing across its network since the start of 2023 [14]. Tires get liquidated through independent dealers and repair shops; electronics, luxury clothing and alcohol are the other stated favourites [15].
The spending response is already committed. Major carriers are working with local police, the FBI and corporate security teams, and investing millions of dollars in monitoring technology, drones, barbed-wire fencing and similar hardening [16]. The opposing capital requirement is a hydraulic bolt cutter [17]. "These gangs can steal tens of thousands worth of goods in minutes, not to mention the damage that can be done to the trains," CSX police chief Sean Douris said [18].
The structural exposure is the railroads' own efficiency work. Trains now stretch as long as three miles and often run with two crew, and criminals have studied the long, predictable stop times as trains approach depots [19][20]. Theft quadrupled from pre-2020 levels after gangs whose drug supply chains were disrupted by lockdowns pivoted to physical goods, concentrating on rail hubs in urban bottlenecks including Chicago, Los Angeles and Memphis [21][22]. CSX names the Gangster Disciples, Crips and Almighty Vice Lords around Memphis; a criminal complaint in US District Court in Arizona alleged Sinaloa-based transnational networks behind a spate of Southwest thefts [23][24].
Watch whether the industry group's 2026 tally moves at all, since surveillance and task forces do not shorten dwell times [6][19]. Watch for carriers disclosing the monitoring spend as an operating line rather than a security footnote [16]. And watch the liability split: if carriers absorb more below the insured layer, the cost lands on operating ratios, not underwriters [9].
Ranked by verification strength, evidence, and original report placement.
In November, three men emerged from a black Kia Soul outside a depot in Memphis, climbed onto CSX Corp. railcars, broke container locks and took more than $20,000 of items including women's clothing destined for a Belk department store, Lazer Blocks toys and two cases of Fre non-alcoholic chardonnay.
The November theft was the type railyards had experienced dozens of times over the past few years, but this time AI-powered cameras were being used to watch the heist, and a task force focused on railroad crime responded.
The team was led by the Tennessee Bureau of Investigation with assistance from federal, state and local agencies, along with the Tennessee National Guard and CSX's own police force, and launched a plane that tracked the vehicle for 11 miles.
Officers moved in a few hours later as one of the perpetrators was unloading boxes of stolen goods, and two other men were arrested separately over the following days, according to police records reviewed by Bloomberg News.
The thieves filled their vehicle and hid the rest of the goods in a nearby wooded area, planning to come back for it later.
Criminals took more than $200 million of goods in 75,000 thefts on US freight rail networks in 2025, according to an industry group that represents CSX, BNSF Railway Co. and Union Pacific Corp., among others.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Documented incidents, single-outlet sourcing
Individual facts are unusually well anchored for a crime feature: a Phoenix federal court filing, an Arizona criminal complaint, police records reviewed by Bloomberg News, an industry-group tally and on-record statements from CSX's police chief and a spokesperson. But the cluster contains exactly one publisher, the industry group and its methodology are unnamed, and headline aggregates (the $200M/75,000 total, the pandemic quadrupling, the 80% Memphis decline) are unaudited and uncorroborated by any second source here.
Funded, quantified deployments at three Class I carriers
This is not pilot talk: dollar spends, fence footage, camera counts, drone types, in-house police headcount and schedule changes are itemized across CSX, Union Pacific and BNSF, and one live intercept shows the detection-to-arrest chain operating end to end. Adoption is scored below the top band because the deployments are company-disclosed, the strongest outcome measure (an 80% regional drop) is unverified, and network-wide coverage beyond named yards is not described.
Mildly overstated by company framing
Slightly positive: the specific loss and spend numbers are grounded, but the causal claims that carry the narrative lean on interested parties. 'Finally catching up with the criminals' and an 80% regional theft decline come from or through the operator whose capex is being justified, with no baseline, no counterfactual and no separation of the AI cameras' contribution from fencing, retiming, task-force staffing or general crime trends. The showcased Memphis haul is also about seven times the average incident by value, making the flagship anecdote atypical of the 75,000-incident base.
Interested sources dominate the causal claims
The principal voices have clear stakes: CSX's police chief and spokesperson are validating a multimillion-dollar security programme, and the industry group supplying the national loss totals represents the carriers themselves. Framing theft as an external gang problem also deflects from the article's own observation that 3-mile trains, two-person crews and predictable depot dwell created the exposure. Countervailing incentives - shippers, insurers, labour, civil-liberties concerns about persistent yard surveillance - are absent from the cluster.
Solid on incidents, weak on trends and efficacy
Confidence is moderate. Discrete, document-backed incidents and disclosed spending figures are reliable enough to act on; the trend magnitude, causal attribution to AI cameras and the 80% improvement claim are single-source and interested, so they should be treated as directional. A second independent publisher or the industry group's underlying methodology would move this materially.
invest
Your Landed Cost Is Being Litigated By Companies With $306,000 Problems1 distinct publisher
security
A North Korean IT worker got hired by a federal agency. Vetting is a security control now.1 distinct publisher
invest
Washington licenses private hacking, and hands the contractor the liability1 distinct publisher
security
Gunra Goes Franchise: Conti's Leaked Code Now Ships With a Builder and an Affiliate Panel2 distinct publishers
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 17, 2026