Project Watershed 250 Red-Teams Texas Water Utilities. The Hard Part Starts in Month Seven
The White House launched a six month, no cost red-teaming pilot across Texas water systems on August 31, with EPA, CISA and Texas Cyber Command. It is a genuinely good program. It also ends, covers one state, and leaves behind findings rather than the ability to see the next intrusion.
- What: The White House launched Project Watershed 250 on August 31, a six month pilot that red-teams Texas water utilities at no cost to them.
- Who: EPA and CISA federally, Texas Cyber Command at state level, with private-sector partners providing intelligence, tooling and training.
- Why now: Officials say it is not a response to a specific breach, while citing attacks on more than thirty Minnesota water systems. The threats are active and documented, and they fall hardest on small utilities with few defences.
- Where we sit: This is the play we have been running with utility districts in Tennessee. The federal program says out loud what we have been telling water operators for two years.
- If you run water: Take the free assessment. Then answer separately: if someone reached your HMI tonight, which log shows it, who reads that log, and how long would it take.
Water is the soft target everyone in this industry has been pointing at for a decade, and it stays soft for an unglamorous reason: the utilities most exposed are the ones least able to pay for defence. A system serving a few thousand connections has operators who know the plant intimately, equipment spanning three decades, and no security staff at all.
Project Watershed 250 is a serious attempt at part of that problem. It is worth understanding precisely what part.
What Was Announced
On Monday, August 31, the administration launched Project Watershed 250, a six month pilot that puts red teams inside Texas water utilities to find weaknesses before an adversary does. National Cyber Director Sean Cairncross framed it as part of a broader push to secure critical infrastructure against determined adversaries.
The structure is worth reading carefully, because the details are what matter to anyone who runs a water system:
- Federal: the Environmental Protection Agency and CISA.
- State: Texas Cyber Command.
- Private: a group of security and AI vendors providing threat intelligence, tooling and training.
- Cost to the utility: nothing.
- Duration: six months, with the intent to expand nationally afterwards.
Officials were explicit that the program is not a reaction to a specific breach, while acknowledging that attacks against more than thirty Minnesota water systems reinforced the urgency. The shape of the problem was stated plainly in the briefing: the threats are active and documented, and they fall hardest on small utilities that have no robust defences to begin with.
This Is the Play We Have Been Running
We read this announcement the way you read a competitor validating your roadmap, which is to say with some satisfaction and no surprise.
RedEye has been doing this work with municipal water and utility districts in Tennessee: finding what is exposed, mapping what actually talks to what, and getting continuous monitoring onto plants that were never going to buy an enterprise SIEM. The federal government has now said out loud, with EPA and CISA behind it, what water operators have been told by their budgets to ignore for a decade. Small systems are targets, the threats are documented, and the utilities least able to fund a defence are the ones carrying the most risk.
So this is not a program we are watching from the outside and picking at. It is the same play, run at federal scale in one state, and it is good that it exists. If you run a Texas water system and can get into the pilot, get into it.
What follows is the part the pilot is not scoped to solve, which is where we spend our time.
An Assessment Is a Photograph, Not a Camera
A red team tells you what was wrong on the days they looked. That is valuable and it is finite. When they leave, the utility holds a report and a list of fixes. What it does not hold is the ability to notice that something is wrong next Tuesday.
That gap is not a criticism of the pilot. It is the difference between assessment and monitoring, and they are separate disciplines with separate budgets. The trouble is that the second one is where the money runs out, and it is the one that has to be paid for every year rather than once.
Three specifics make it sharper for water:
- Six months ends. The pilot has a defined stop. Intrusions do not schedule themselves around grant windows, and the quiet, patient access that matters in OT is exactly the kind a point-in-time assessment is least likely to catch.
- One state, some systems. There are roughly fifty thousand community water systems in the United States. Texas is a serious start and it is a small fraction of the problem.
- Small utilities stay small. The utilities identified as most exposed are the ones with a couple of operators, no security staff and no analyst on call at 2am. A better report does not change that. Something has to watch the plant when nobody is looking at it.
What Small Utilities Actually Run Into
We sell into municipal water and utility districts, so this is not theoretical. The pattern is consistent.
The operators are competent and stretched. The OT network is a mix of equipment from three decades, some of which cannot be patched and none of which can be taken offline for a maintenance window that suits a vendor. There is usually no SOC. There is frequently no full time IT person, let alone a security one.
And when these utilities do look at monitoring, they discover the economics were designed for someone else. Traditional platforms price on ingest, so the more of the plant you watch, the more you pay, and the rational response is to collect less. That is precisely backwards for OT, where the signal that matters is a small anomaly inside a large volume of ordinary process traffic. A utility that samples its historian to control cost has bought a monitoring system that is blind in the exact place it needed to see.
That is the constraint we built against. Storage economics are not a marketing line for this buyer, they are what decides whether the plant is monitored at all.
What We Have Been Building For This
Caver is our security data platform, and the OT work in it exists because of conversations with utilities like these rather than a market study.
- Zone and conduit visibility. Segmentation as a picture of what actually talks to what, rather than a diagram in a binder that stopped being true in 2019.
- Process integrity. Learned operating envelopes instead of static thresholds, because a pump that is behaving strangely for this plant, this season and this load is not a value you can hardcode.
- Safety and availability. The screen an operations manager actually wants, which is usually not the one a security vendor builds first.
- Compliance evidence. Mapped to what proves the control, so an audit is a query rather than a fortnight of screenshots.
- Vendor and remote access. Who reached in, from where, to touch what. In water this is frequently the whole story.
It runs on storage the utility already owns, keeps full fidelity rather than sampled data, and is queryable in the language the operator already knows. There are two ways to adopt it: migrate fully, or run it in parallel with what you have until you trust it.
Month Seven Is the Whole Problem
A pilot proves a thesis. It does not staff a plant.
When the six months are up, the utility is holding a report and a list of remediations, and it is back to whatever it had before, which for most small systems is nothing watching the OT network at all. The same is true today for every water system outside Texas, which is nearly all of them.
That is the question worth asking now rather than in March: when the assessment is finished and the team has moved on, what is left running that would notice an intruder in your SCADA network on an ordinary Tuesday? If the answer is nothing, that gap is not closed by another assessment. It is a monitoring problem with a budget problem attached, and it is the one we built for.
If You Run a Water System
Three things worth doing this month, regardless of who you buy anything from:
- Apply to the pilot if you are in Texas. It is free and it is real. There is no reason not to.
- Write down what you would actually see. If someone reached your HMI tonight, which log would show it, who reads that log, and how long would it take. For most utilities the answer is nobody and never, and that is a finding you can act on without buying anything.
- Ask what your monitoring costs at full fidelity. Not sampled, not filtered down to fit a licence. If the number only works because you are collecting a fraction of the plant, you have found the same constraint everyone else in this sector has.
We are in Tennessee and we work with utility districts on exactly this. If you want a second opinion on what your plant would and would not see, that conversation is free and does not come with a pipeline.