Gene and I spent the first half hour of episode 182 talking AI, and I figured it was worth writing down where I’ve landed, at least for now. Short version: AI is the most useful tool I’ve picked up in years, and I still won’t give it a single password.

Here’s what I use it for. I finally have my first truly recurring task running on my Grok bot. Every Monday it goes out and hunts down RFPs related to my business. It pulls the opportunity, the issuing entity, the title, the date it was posted, the date it’s due, scope notes, and whether it’s open or limited. All of it lands in a nice spreadsheet.

The interesting part is how the work is split. Grok only handles the internet-facing stuff, the scraping. It hands what it finds to a local Qwen model on my own hardware through an MCP bridge, and the local model does the analysis, summarizing, and formatting. I could have Grok do the whole thing. However, I already own the hardware, it’s cheaper, and getting the two to work together was an experiment. Now I have a sales resource.

Note: Grok has none of my passwords and it never will. I’m not logging Grok or any other AI into my Servers to troubleshoot. I’ll ask it for troubleshooting steps and do the work myself.

Executive are bragging that their AI reads their email, sets up their meetings, runs their computer, and manages their bank account. That’s nuts! If you read [my post on the water systems](https://namedben.com/water-systems-hit-across-at-least-7-states/), it’s the same lesson. Don’t leave the keys sitting out where anything can grab them.

And yes, even a locked-down bot can bite you. Gene set his AI up on Signal so it would only listen to him. So of course I started sending it SQL injection, drop table commands, the usual. Gene replied to me, quoting my command back to tell me it wasn’t going to work… and the bot answered, because now the command came from the one guy it listens to I’m reminded of the Warren G. Harding quote: his enemies he could handle, it was his friends that kept him up at night.

Which brings me to my rule: you can’t ask it about something you don’t already know enough about. Well, you can, you just won’t know if the answer is right.

So : 1) let it do the grunt work,

2) verify everything it hands you,

3) never hand it credentials.

And Yes I used Grok to write this post but hey I edited it!

The whole conversation is in episode 182, “When Machines Can Copy Software From A Screenshot.” [Give it a listen](https://podcastindexrobot.com/open/5770690/61136724489).

More episodes at [justtwogoodoldboys.com](http://justtwogoodoldboys.com).

I’ve been watching this one unfold since Minnesota first reported it last week. More than 30 municipal water and wastewater systems targeted there on the 26th and 27th. Now the FBI and EPA say water utilities in at least seven states have reported incidents, and some of that activity degraded operations. Michigan has confirmed nine systems saw hostile activity. Georgia and South Dakota (Rapid City lift station) have also come forward. New Jersey is in the mix too. The rest haven’t all been named publicly yet.

The playbook is the same one we’ve been warning about for years: internet-facing PLCs. Mostly Rockwell/Allen-Bradley from what I’m seeing, with others getting touched as well. Actors log in, change the IP address, change the password, and lock the operators out. In some places that meant loss of pressure, flooding, boil-water notices, and a forced switch to manual operations. No confirmed contamination or public health impacts so far. The systems so far have kept delivering safe water. That part is good news.

Intelligence folks are pointing at Iranian-linked actors, I don’t care about attribution. I care about the attack, and how to beat my opponent. Given the current state of things with Iran, that tracks with previous OT targeting patterns. Official attribution is still cautious, which is fine. It doesn’t change the fact that these devices should never have been sitting on the open internet in the first place.

I’ve said it before and I’ll say it again: the water sector is full of small utilities with tiny budgets, aging equipment, and staff who are already stretched thin just keeping the pumps running. NERC CIP-style rules never applied here the way they do in electric. PLCs were never designed to face the public internet, yet there they sit with default or weak credentials. This isn’t sophisticated zero-day stuff. It’s opportunistic access against systems that were left exposed.

CISA’s advice is the correct one: pull those controllers offline or put them behind proper segmentation and access controls yesterday. If your HMI or PLC is reachable from the public internet, you are already behind. Manual operations are better than remote compromise.

Said differently if you have assets on Shodan, you are fucked and have been. DO YOUR JOB!

I know this space. I’ve sat through the hearings. I’ve watched operators scramble. The fact that most systems recovered quickly and water stayed safe is a credit to the people on the ground. The fact that it was this easy is a failure of priorities and basic hygiene.

Leave the critical systems alone. Secure them. The men who just want to keep the water flowing shouldn’t have to fight off remote password changes while the rest of us argue about who to blame.

“Extremism in defense of liberty is no vice. Moderation in pursuit of justice is no virtue.” – Goldwater

Jack was my first employer and friend. I have nothing but good things to say about the man. Please hear his words.

https://www.youtube.com/watch?v=NUASH-B8ybA&list=PLvbBokkbXMa2LT2u5pX9Ar-qe6psKnpRK

Well Since Gene played part of it on Unrelenting. Here is the new Theme AI song for our podcast.

https://time.com/6821425/israel-embassy-air-force-protest-fire-self-immolation-aaron-bushnell-latest-updates/

Some commemorators have called the Airman who decided to end his own life in protest as “obviously mentally ill”. The Video linked above is from the Vietnam war when a Buddhist monk did the same act as a protest. I cannot recall ever reading a history text that called him “mentally ill” for his actions. While I do not agree with the Airman’s assessment that the current conflict warranted his actions; there is currently nothing in current geopolitics that would drive me to the same actions. I do, however, see his actions as that of a desperate man who wanted to drive awareness to his views in a world that rarely listens. Often in the west we hear “fuck around and find out” this is the same thing without the threat of violence (at least implied) to others. This act of self harm did what he wanted it to do. It made the news and made (at least me) question why someone would be willing to do this.

The question I pose to you is what is your “line in the sand” where events become so intolerable that you will not continue to exist as you do today in order to change the way the world sees the situation. Some who have gone down this path have (most commonly) chosen violence instead of a peaceful, though deadly protest.