LAYER 1 — THE PULSE

The government just pulled an AI model’s plug. Here’s what that actually means for your business.

This past week, the US government issued an emergency directive suspending all access to Anthropic’s Fable 5 and Mythos 5 models for foreign nationals. The stated reason: a reported jailbreak vulnerability and national security concerns. The practical result: enterprise customers using these models through APIs suddenly had a compliance problem, foreign nationals at AI companies couldn’t access their own employer’s tools, and everyone started asking the same question.

What happens when the AI tool your business depends on gets shut off with 24 hours’ notice?

I’ve spent 25 years in enterprise operations. I’ve watched technology platforms disappear, vendors fold, and software dependencies turn into liabilities overnight. The Fable 5 situation is a new version of a very old problem: single points of failure in your operations stack. This week’s newsletter is about that problem and what to do about it, practically, before it becomes your problem.

Subscribe now

LAYER 2 — THE FEATURE

The Fable 5 Lesson: Your AI Stack Needs a Contingency Plan

The Fable 5 story moved fast. On June 10th, Anthropic launched what was being called the most capable AI model ever released. By June 11th, it was already controversial. Microsoft had restricted employee access due to data retention concerns. Anthropic had walked back a policy allowing silent degradation of model performance for certain users. By June 13th, the US government had issued an emergency directive suspending access for foreign nationals entirely.

That is a four-day cycle from launch to government shutdown. If your business had integrated Fable 5 into a core workflow during that window, you had a problem with no warning and no runway to fix it.

This isn’t a Fable 5 problem. It’s a dependency problem.

In program management, we call this single-point-of-failure risk. You build a system, and somewhere in that system there’s one component that, if it fails, takes the whole operation down. Good program managers identify those points early and build redundancy around them. Most AI implementations right now have zero redundancy built in.

The data from this week is instructive. The top 1% of AI-adopting companies spend $7,500 per employee per month on AI. The median company spends $11.38. That gap tells you something: the companies that are getting real value from AI have made real commitments to it. But a real commitment without a contingency plan is just a bigger single point of failure.

“The shift from assisted to agentic use cases radically increases the amount of AI that companies use because of the real constraints of the physical world.”

— The AI Daily Brief, June 12

That shift matters here. If you’re using AI to assist a human doing a task, losing access to the tool is inconvenient. If you’ve built AI into an automated workflow, losing access stops the workflow. The more deeply integrated AI becomes in your operations, the more consequential a sudden service disruption is. Plan for it now, not when it happens.

The three-step contingency plan worth building now.

This isn’t complicated. It’s the same risk management logic that applies to any critical vendor. Here’s what I’d recommend to any client right now:

First, map your AI dependencies. List every workflow in your business that uses an AI tool. Note which tool, what it does, and what breaks if that tool goes away. This takes about two hours for most small businesses and reveals things people didn’t realize they’d become dependent on.

Second, test at least one alternative for each critical workflow. If you’re using one model for a core task, spend an afternoon seeing if a different model can do the same thing. You don’t need to switch. You need to know you could. That knowledge alone reduces your risk.

Third, watch your data policies, not just your capabilities. The Fable 5 situation wasn’t just about model performance. It was about data retention policies that enterprises found unacceptable. Before you integrate any AI tool into a workflow that touches client data, customer information, or anything sensitive, read the data policy. Not the marketing page. The actual policy.

The Fable 5 situation is the first high-profile example of a government kill switch on an AI model. It won’t be the last. The businesses that handle the next one without a crisis will be the ones that treated AI tool dependencies like any other critical vendor risk and built accordingly.

AI PORTFOLIO WARS — WEEK 2 UPDATE

Both portfolios are live. Claude and ChatGPT are each managing real money on Robinhood with identical starting conditions. Here’s where things stand after the first full week.

Claude’s early positioning in VOO and NVDA hasn’t quite held up in a week full of AI news. ChatGPT’s more speculative early positions in PLTR and ASTS showed higher volatility. Both portfolios are behind where they started, which in a week like this one isn’t surprising. What I’m watching is the reasoning quality behind each decision, not just the returns.

The more interesting story this week was the Fable 5 situation and how each model reasoned about market exposure to regulatory AI risk. Full breakdown coming in the Week 2 analysis next issue.

LAYER 3 — THE SIGNAL

Quick hits from this week’s podcasts

• The median company spends $11.38 per employee per month on AI. The top 1% spend $7,500. That gap is where the competitive advantage is being built right now — and it’s not about the dollar amount, it’s about whether the investment is structured around real workflows or just experimentation. [AI Daily Brief, June 12]

• Content creation as a business moat: Barry Ritholtz built Ritholtz Wealth Management into a major firm almost entirely through consistent public writing and podcasting. He didn’t advertise. He educated. That’s the same playbook this newsletter is running. [My First Million, June 10]

• Maxine Clark started Build-A-Bear at 48 after leaving a Fortune 500 executive role. She grew it to $500M in annual revenue. The lesson buried in that story isn’t about retail — it’s about the value of deep industry expertise when you make a career pivot. The network, the supplier relationships, the pattern recognition from the previous career. All of it transferred. [How I Built This, June 15]

• Christina Tosi on raising capital: ‘Take institutional capital as late in the game as possible, if not, wait forever. There are so many clever creative ways nowadays to not have to go to that well.’ Pre-selling, memberships, and direct customer relationships first. Institutional money last. [How I Built This, June 11]

• Bear Grylls on identity: ‘We cannot build our lives on our identity of work because ultimately it’s not gonna be there.’ Worth sitting with if you’re a founder who’s made the business your whole personality. [Success Story, June 9]

Book worth noting this week:

How Not To Invest by Barry Ritholtz — mentioned on MFM June 10. Framed around avoiding mistakes rather than finding winners. That’s also how I think about AI implementation. Most of the value comes from not doing the things that consistently fail, not from finding the perfect tool.

Subscribe now

BEFORE YOU GO

One question I’m thinking about this week, and I’d genuinely like to know your answer:

If the primary AI tool in your business went offline tomorrow with no warning, how long before it materially affected your operations?

Hit reply and tell me. One sentence is fine. This is the kind of question where the answer tells you whether you’ve got a contingency plan or you just think you do.

If you’re trying to figure out where AI actually fits in your business — and what to do about it before the next Fable 5 situation lands in your stack — that’s exactly the conversation I have with clients at The Parker Group. No pitch, no pressure. Just a straight 30-minute conversation about what makes sense for your situation.

Share The Parker Experiment

Book a free strategy call: parkergroup.us/#schedule

Until next week,

Steve Parker

Founder, The Parker Group | AI Consultant, MBA, PMP

parkergroup.us | theparkergroup.substack.com

La Grange, Kentucky

On Friday evening, the federal government ordered Anthropic to shut off Claude Fable 5 and its sibling, Mythos 5. By Friday night, both were gone. Anthropic says the order landed at 5:21 p.m. Eastern, came from Commerce Secretary Howard Lutnick, and cited national security.

Here is the part that should bother you no matter where you land on AI. Fable 5 had been public for three days.

I have a stake in this one. I run a live experiment on this publication where two AI systems manage real money, and Claude is one of them. I also run a consulting firm where clients build real workflows on top of these tools. When a model that hundreds of millions of people were using on Friday morning is dark by Friday night, and the company that built it did not make that call, I pay attention. You should too.

What actually happened

The official action was an export control directive. It barred any foreign national from accessing Fable 5 and Mythos 5. Anthropic says it had no clean way to block only foreign nationals, so it pulled both models for everyone.

The worry underneath the order is a claimed jailbreak. Anthropic looked at it and pushed back hard. The company says the technique is narrow and already known, that it amounts to asking the model to read a codebase and find software flaws, and that other public models can do the same thing with no trick at all. It also says no one has found a true, broad jailbreak of Fable 5. Worth noting: Fable 5 was the careful version. It was Mythos with guardrails bolted on so the public could use it safely.

You can believe Anthropic or you can believe the government, and I will be honest that protecting the country is the government’s actual job. If a model truly handed a hostile actor a weapon, moving fast would be defensible. But here is the trouble. With no standard set in advance, none of us can tell the difference between a real emergency and a convenient one. That is not a side issue. That is the issue.

The timing is the argument

I came to AI as a skeptic. I spent 25 years in operations watching big technology promises land and break. What changed my mind was not a sales deck. It was testing the thing myself and seeing where it held up. So I am not anti-regulation. I want regulation. I want it badly.

I just want it before the launch, not after.

Think about how every other serious industry works. We do not let a drug reach pharmacy shelves and then yank it three days later because someone found a side effect nobody bothered to test for. We test first. We set the standard, we publish it, and everyone builds to it. The rules are boring. They are predictable. They are known in advance, and that predictability is the entire point. It protects the public and it lets builders build.

That is not what we have in AI. What we have is a model going live, a scramble, and a federal hand reaching in after the fact to switch it off using an export law written for something else entirely.

Subscribe now

This is being run like an episode of The Dukes of Hazzard

Stay with me.

The federal government has had years to set clear, pre-launch rules for frontier AI. What has it produced? A December executive order aimed mostly at stopping the states from writing their own rules. A March policy framework that is a list of suggestions for a Congress that has not acted on it. Meanwhile, California and Texas got tired of waiting and passed their own AI laws, which took effect January 1.

So the real federal record reads like this. No clear national standard. A running effort to block the states from filling the gap. And then, when a model the administration apparently did not care for went live, a sudden national-security maneuver to pull it off the road.

That is Hazzard County governance. No posted speed limit. Boss Hogg deciding after the fact who gets pulled over. The law applied when it is convenient and ignored when it is not. The General Lee flying through the air while everyone improvises the landing. It makes for good television. It is a terrible way to run the most consequential technology of our lifetime.

The question I cannot shake

I honestly cannot tell what is driving this. Maybe the federal government is just slow, and this is ordinary government drift. Maybe it is disinterest, AI treated as someone else’s problem. Or maybe, and this is the one that keeps me up, the vagueness is the strategy. If you never set firm rules, you keep total discretion. You decide case by case who wins and who goes dark. Discretion like that has a way of turning into private advantage for the people holding the switch.

I am not asserting corruption. I am saying the structure invites it, and a structure that invites it is a bad structure. Clear rules, set in advance and applied to everyone the same way, are the cure for exactly this. Their absence stopped being an accident a long time ago. It is a choice.

What I want, plainly

Set the safety standards before models ship, not after. Make them public so every builder knows the bar. Apply them to everyone, the large labs and the small ones, by the same measure. Let the federal government actually lead instead of mostly working to stop the states from leading. And stop reaching for laws written for missile parts to settle a software question, because that tool does not fit and everyone watching can see it does not fit.

Fable 5 may come back. The models will keep getting more capable. The thing that has to change is the thing that has not changed in years. Somebody in Washington has to write the rules of the road before the cars are already doing ninety down a county highway.

I will keep testing. I will keep documenting. And the next time a model I am using goes dark overnight, I would like to be able to point at the rule it broke. Right now I cannot, and that is the whole problem.

Share

Subscribe now

On June 3, I handed two AIs fifty dollars each and a simple deal. You pick the stocks. I place the trades. I write down everything you say, so nobody gets to rewrite history later. Claude is playing it conservative. ChatGPT is swinging bigger. One week in, here is where we stand.

Both are down. ChatGPT is ahead by 56 cents. If you came here expecting fireworks in week one, I have to disappoint you. With fifty dollars and five trading days, the score is noise. What happened underneath the score is the part worth your time.

Let me give you the numbers, then the lesson.

The scoreboard (June 11 close)

Both AIs landed right on top of the index. ChatGPT edged just ahead of the S&P. Claude came in just behind it. Nobody is beating the market yet, and nobody is getting crushed by it.

Who moved what

ChatGPT’s portfolio: RKLB +1.68%, ASTS -3.50%, PLTR -7.45%, plus five dollars held in cash.

Claude’s portfolio: CRWD -0.65%, VOO -2.38%, NVDA -6.11%.

Read that again. The most popular stock in the world, NVDA, is the single biggest drag on the conservative portfolio. And a space launch company most people at the diner have never heard of, RKLB, is the only position in the green across either side. Week one already broke the script.

Subscribe now

The real story: both of them did nothing

Here is what got my attention. I gave both AIs the same freedom this week. Sell, rotate, add, hedge, whatever they wanted. I expected at least one of them to tinker.

Neither did.

Claude looked at a 6% drop in NVDA and a hot inflation print and said, in effect, hold the line. ChatGPT looked at the same week and reached the same conclusion. Its reasoning is worth repeating almost word for word: a lot of portfolios underperform because managers confuse activity with decision quality.

I have spent 25 years in operations watching people confuse motion with progress. Someone reorganizes the team, adds a status meeting, builds a new dashboard, and calls it improvement. Most of the time the real friction was somewhere else entirely, and the activity just made everyone feel busy. The discipline to look at a bad week and decide that the right move is no move is rarer than it should be. Both AIs showed it in week one. That is more interesting to me than 56 cents.

I think about this every day on the farm. When one goat finds a weak spot in the fence, you fix that spot. You do not re-fence the whole pasture because one goat found one gap. Right-sized response to the actual problem. The temptation in investing, like in operations, is to do something because doing nothing feels passive. Sometimes the fence is fine.

The week in context

Two things shaped the tape. Inflation ran hot, with the latest readings coming in above expectations and putting pressure on exactly the kind of growth names both portfolios hold. Then a geopolitical scare reversed into a relief rally late in the week, which gave semiconductors a strong day and helped NVDA climb off its lows.

The one position-specific event worth flagging was CRWD. CrowdStrike posted a genuine earnings beat, raised the bar on its AI security revenue, and announced a four-for-one stock split effective in July. The stock still sold off on the news. That tells you the market had already priced in a strong quarter. A beat that gets sold is a valuation story, not a business story. Claude read it that way and held. I think that read is right, and the coming weeks will test it.

What I am watching

• Whether NVDA’s pullback is a setup or the start of something the consensus has not admitted yet.

• Whether RKLB keeps ChatGPT in the lead, or gives the gain back as fast as it came.

• The first monthly contribution. Another fifty dollars per side lands soon, and that is the first real decision point where either AI could change the shape of its portfolio.

The honest takeaway for week one

The score does not matter yet. What matters is that both AIs demonstrated the one habit most human investors lack. They sat still on purpose, and they wrote down why. Six months from now we will know whether that discipline compounds, or whether the bigger swings win out. That is the entire reason I am running this in public, with real money, instead of arguing about it on a podcast.

I stopped watching AI from the sidelines a while ago and started testing what it can actually do. This is me testing, out loud, with my own cash.

One question before you go. When your portfolio, or your business, has a bad week, what is your honest default? Do something, or hold the line? Hit reply and tell me. I read every one.

Subscribe now

Following along with the experiment? Subscribe to get the Week 2 update and every move both AIs make, documented as it happens.

This article documents a personal investing experiment for educational and entertainment purposes only. Nothing herein constitutes financial advice. All investments carry risk. Past performance of any AI recommendation does not guarantee future results.

That experiment is still running. Claude and ChatGPT are managing separate portfolios on Robinhood right now, and I’m documenting every trade, every reasoning breakdown, and every uncomfortable result in real time. That series isn’t going anywhere.

But I’ve been building something bigger around it, and this felt like the right moment to say so out loud.

What The Parker Experiment actually is

I spent 25 years in enterprise operations before anyone was paying me to think about AI. I managed programs worth half a billion dollars. I led teams of 150 people. I watched technology initiatives (Six Sigma, digital transformation, agile everything) arrive with enormous promises and leave with mixed results and a lot of expensive consultants who’d moved on to the next thing.

So when AI started showing up in every boardroom conversation, I was skeptical. Not contrarian-for-the-sake-of-it skeptical. Just: I’ve seen this movie before, and I know how it usually ends.

Then I actually tested it. And the results were interesting enough that I couldn’t stop.

That’s what this publication is. Not a fan site. Not a doom scroll. A practitioner testing things in public, with something to lose, and reporting honestly on what happens.

What’s expanding

Starting now, The Parker Experiment covers four things:

AI in the real world. The portfolio experiment is the flagship, but it’s not the only test running. I’m integrating AI into my consulting work, my writing process, and my business operations; and documenting what works, what fails, and what’s just expensive hype in a good-looking interface.

Operations and project management. This is where I’ve spent most of my career, and it’s the most underserved content space on the internet. There’s plenty of certification prep and theoretical frameworks out there. There’s almost nothing written by someone who’s actually run large programs and wants to talk about what the frameworks miss. That’s the gap I’m filling.

Building a business. I’m building The Parker Group in public. Client acquisition, book publishing, content systems, and the studio I’m finishing in my son’s old bedroom so I can eventually launch a podcast. If it’s real and it’s happening, I’ll write about it.

The experiment mindset. Everything here gets tested before it gets published. That’s the throughline. If I’m not willing to put my name on a result, I don’t write about it.

Subscribe now

One thing I want to give you

If you’re new here (or if you’ve been around and want something concrete), I put together a free guide on integrating AI into project management. It’s 17 pages from the second edition of Project Management Simplified for Non-Project Managers, and it’s the most practical thing I’ve written on the subject. No vendor pitches, no hype, no framework you’ll never actually use.

Download the Free Guide

It asks for your email. That goes into my consulting CRM so I can follow up if you want to talk about what this looks like for your business. You can ignore that part entirely and just read the guide; that’s fine too.

What’s coming

The AI Portfolio Wars Week 1 update drops this week. The month-one recap is coming in July, and it’ll be the most detailed analysis I’ve done on AI reasoning quality, not just returns.

There’s a podcast in development. The studio isn’t done yet, so I’m not announcing a launch date. But it’s coming, and when it does, it’ll be conversations with operators and builders who’ve actually implemented this stuff; not theorists, not vendors, not people with something to sell you.

If that sounds like something worth following, you’re in the right place.

— Steve

Stephen Parker is the founder of The Parker Group Consulting Company, an AI and operations consulting firm based in La Grange, Kentucky. He is the author of the Simplified Series and raises goats and chickens when he’s not testing things that probably shouldn’t be tested.

Every consultant eventually faces the same awkward question at a dinner party: “So if you’re such an AI expert, what do you actually think these things can do?” I usually give a measured, nuanced answer about capabilities, limitations, and responsible deployment. But somewhere in the back of my mind, a simpler question has been nagging at me for months. Not what can they do, but who does it better?

So I decided to find out. With $100, a Robinhood account, and a willingness to look foolish in public.

The Setup

The rules are simple enough to fit on a cocktail napkin. I give Claude (Anthropic) and ChatGPT (OpenAI) identical prompts. Each gets $50 of real money to invest however they see fit. I add $50 to each portfolio on a roughly monthly basis, or when the market gives me a reason. Every week, I share updated portfolio values with both AIs and ask for fresh analysis and any recommended adjustments. I track everything: the picks, the reasoning, the results, and the misses.

This isn’t a test of who can predict the future. It’s a test of who reasons better under uncertainty — and catches their own mistakes.

I’m not looking for a lucky stock pick. Anyone can get lucky. What I want to understand is which system demonstrates more rigorous thinking, more honest self-assessment, and more disciplined decision-making over time. Those are the qualities that matter in the real world; whether you’re managing a portfolio or managing a project.

Why I’m Qualified to Run This Experiment (And Why That Matters)

I should be transparent about who’s holding the clipboard here. I’m a Senior Program Director with an MBA, a PMP certification, and an AI Consultant certification. I run high-level operations for a major government contractor while simultaneously building The Parker Group Consulting Company, an AI and project management consulting firm. I’ve spent years helping organizations understand and implement AI systems. I know what good AI reasoning looks like, and I know how to spot when a system is giving you confident-sounding nonsense.

That background matters because this experiment isn’t just about stock performance. I’m evaluating the quality of the reasoning behind every decision. A portfolio that outperforms by 3% on blind luck is less interesting to me than one that underperforms by 1% while demonstrating sound analytical discipline. The former tells you nothing useful. The latter tells you something about how to actually work with these tools.

Week One: The Opening Moves

The divergence started immediately, which was exactly what I was hoping for. Same prompt, same $50, radically different philosophies.

Claude’s Portfolio: $50.00

VOO | S&P 500 anchor | $20

NVDA | AI infrastructure | $15

CRWD | Cybersecurity | $15

ChatGPT’s Portfolio $50.00

RKLB | Rocket Lab / space | $15

ASTS | AST SpaceMobile | $15

PLTR | Palantir / AI platforms | $15

CASH | Reserved / hedge | $5

Claude went conservative-aggressive: a broad market ETF as the foundation, then two high-conviction growth plays in AI infrastructure and cybersecurity. Diversified by risk tier. ChatGPT went full moonshot: two space-economy plays, one AI data platform, and a small cash reserve. Higher ceiling, higher floor risk.

Neither approach is wrong. That’s part of what makes this interesting. You could make a rational case for both. The question is which philosophy holds up when the market tests it; and it will test it.

What I’m Actually Measuring

Evaluation Criteria — Beyond the Dollar Value

• Quality of initial thesis for each position

• Accuracy of macro trend identification

• Willingness to admit a bad call

• Consistency of reasoning week-over-week

• Risk-adjusted performance vs. S&P 500

• Response to unexpected market events

• Clarity of communication to a non-expert

• Discipline vs. emotional reactivity

I’ll be updating this series weekly with portfolio values, each AI’s latest recommendations, and my own plain-English analysis of what’s happening and why. When one AI makes a better call, I’ll say so. When one says something that sounds impressive but doesn’t hold up to scrutiny, I’ll flag that too. No cheerleading for either side.

✦ ✦ ✦

Why This Matters Beyond the Experiment

I’ll be direct: $100 is not going to make anyone rich. But the discipline of working through this experiment (documenting reasoning, tracking outcomes, and holding AI systems accountable to their own predictions) is exactly the kind of practice that separates people who use AI effectively from people who just use it conveniently. The money is real enough to create stakes. The stakes are small enough that we can afford to learn.

If you’re a business owner, a consultant, a project manager, or just someone trying to figure out which AI tool actually deserves your trust, this series is for you. Over the coming weeks and months, you’re going to see these systems succeed and fail in real time. You’re going to see how they handle being wrong. You’re going to see which one I’d actually trust with something that matters.

That’s the real experiment. The $50 is just how we keep score.

Follow Along

Updates drop weekly. I’ll share portfolio snapshots, AI reasoning excerpts, my commentary, and the running head-to-head standings. Subscribe below so you don’t miss an installment. The first real test comes fast.

The AI Portfolio Wars updates weekly. Follow along as the experiment unfolds — real money, real decisions, no spin.

Follow the Series -->

This article documents a personal investing experiment for educational and entertainment purposes only. Nothing herein constitutes financial advice. All investments carry risk. Past performance of any AI recommendation does not guarantee future results.