He's a senior cybersecurity threat researcher known for satirical takes on corporate governance.

Peter Girnus 🦅 @gothburz
24 Feb 2026 · 8:30 PM UTC
https://nitter.net/gothburz/status/2026394374028882107

Last year the board created a new position.

Chief AI Officer.

They gave it to me.

I had no qualifications.

I had a LinkedIn post about prompt engineering that got 4,000 likes. Someone on
the nominating committee saw it. They said I had "a demonstrated fluency in the
space."

The space.

Nobody defines the space. The space is whatever the quarterly earnings call
needs it to be.

My first act was to commission a roadmap.

McKinsey built it. $4.2 million. Fourteen weeks. The roadmap recommended
building a Center of Excellence. The Center of Excellence's first deliverable
was a second roadmap.

I approved both roadmaps.

The second roadmap recommended retaining Accenture.

Accenture recommended retaining BCG.

BCG recommended a "maturity assessment."

The maturity assessment costs $1.8 million and measures how mature your AI
program is on a scale from 1 to 5. We scored a 1.4. BCG said this was "expected
for organizations at the early stages of their transformation journey."

The transformation journey.

The journey costs $480 an hour. The tools cost $30 a month. I pay sixteen times
more for the people explaining the tools than I pay for the tools. This is
called "strategic implementation."

I built the Center of Excellence. Eleven people. A director, two program
managers, three analysts, a communications specialist, a change management lead,
two vendor liaisons, and an AI Ethics Advisor.

The AI Ethics Advisor's job is to attend meetings where no one asks ethical
questions.

She has attended all of them.

The Center of Excellence's primary output is a biweekly newsletter called "AI
Horizons." It goes to all 14,000 employees. The open rate is 6%. I know this
because the analytics dashboard for the newsletter is the most functional AI
tool we've deployed.

I reported to the board that we had "reached 14,000 stakeholders with our AI
literacy initiative."

That is technically true.

Reaching is not the same as reading. But reaching is a metric. Reading is not.

In September I rolled out an AI-powered meeting summarization tool. It replaced
Janet. Janet took notes by hand. Janet was accurate. Janet cost $58,000 a year.

The AI tool costs $340,000 a year in licensing, integration, and cloud
compute. It summarized a board meeting by attributing the CFO's layoff
announcement to the head of diversity. It summarized a product review by
recommending we "sunset the customer." We have not sunsetted any customers. We
sunsetted Janet.

Janet was a contractor. Contractors don't count in the headcount reduction. She
counts in the AI displacement metric. I created that metric. It goes up and to
the right.

In October I launched AskHR, an AI chatbot for employee questions. It was
supposed to reduce HR ticket volume by 40%.

In the first week, it told an employee his paternity leave was fourteen
months. It was fourteen days. The employee told his wife. His wife told her
employer. HR told me.

I told the board the chatbot had "achieved a 40% deflection rate."

Deflection means the employee stopped asking.

It does not mean the employee got an answer.

Forty percent of people who interact with AskHR give up. I call this
deflection. Customer Success calls this resolution. The dashboard calls this a
closed ticket. The employee calls it nothing, because nobody asks the employee.

In November, Goldman Sachs published a report. It said AI had contributed
"basically zero" to GDP despite $650 billion in infrastructure
spending. Basically zero. I forwarded the report to the CEO with the subject
line: "This is why we need to ACCELERATE."

He agreed.

We accelerated.

We are now spending $28 million annually on AI initiatives. The initiatives are:
one newsletter with a 6% open rate, one chatbot that hallucinates paternity
leave, one meeting summarizer that recommends sunsetting customers, one AI
Ethics Advisor who has never been asked an ethical question, and one maturity
assessment that says we're a 1.4.

Only 25% of enterprises report that generative AI has transformed their
business. I know this because Deloitte surveyed 3,235 leaders and put it in a
PDF. I put the PDF in the board deck. I highlighted the 25% and wrote "massive
upside remaining." The board loved that. Massive upside remaining means we
haven't seen results yet and we're choosing to frame that as an opportunity
rather than an indictment.

Only 25% believe their IT infrastructure supports enterprise-wide AI. I know
this because IBM surveyed 400 leaders and put it in a different PDF. I put both
PDFs in the same slide. I titled the slide "The Readiness Gap." I did not title
it "We Are Not Ready."

Forty-four percent of organizations have no measurement framework for generative
AI. I am in the 44%. I have a dashboard. A dashboard is not a measurement
framework. A dashboard is a picture of numbers. The numbers go up because I
designed them to go up. Adoption is measured by logins. A login is counted when
the application opens. It opens automatically when you turn on your laptop.

Every employee in the company logs in to the AI platform every morning.

None of them meant to.

I reported this as "98% daily active engagement." The board gave me a standing
ovation. Standing ovations are not a measurement framework either but they are
the only metric that determines my bonus.

In February, OpenAI's COO said - and I need you to hear this - "we have not yet
really seen AI penetrate enterprise business processes."

He said this out loud.

At a conference.

He runs OpenAI's enterprise division. He said the product has not penetrated the
market he is selling it to.

The next day, OpenAI announced "Frontier Alliances" - multi-year partnerships
with McKinsey, BCG, Accenture, and Capgemini.

The AI company hired the consultants.

The consultants had already been hired by us to implement the AI company's
product.

The AI company is now paying the consultants to help the consultants help us use
the AI that the AI company built. This is not a loop. This is a business
model. The business model has a name. The name is "ecosystem."

BCG's CEO said, and this is a direct quote: "AI alone does not drive
transformation. It must be linked to strategy."

Strategy is the thing BCG sells.

The same week, Anthropic's head of product said: "2025 was meant to be the year
agents transformed the enterprise, but the hype turned out to be mostly
premature."

Premature.

That is the word you use when you want to say "we were wrong" without accepting
that anyone was harmed.

108,435 people lost their jobs in January. That's the highest monthly total
since 2009. Some of those people lost their jobs because companies like mine
said AI would replace them. The AI has not replaced them. The AI has replaced
their budget line. Their work is now done by the remaining employees, who are
doing two jobs for one salary, which we call "efficiency gains."

Efficiency gains go in the board deck.

The remaining employees got raises this year. Flat raises. Same percentage for
everyone. The industry calls this "peanut butter." You spread it thin and even
so no one complains loudly enough to matter. Forty-four percent of companies are
doing peanut butter raises this year. It is the same tactic we used after the
2008 recession. We are running the 2008 playbook while claiming to be in the
future.

I attended the World Economic Forum in January. I was on a panel called
"Responsible AI at Scale." There were four of us on the panel. None of us could
define what our AI platforms measurably do. All of us used the word
"transformative." One of us used it four times in a single answer.

That one was me.

I have created an AI Readiness Score. The score is calculated by multiplying the
number of AI tools licensed by the number of departments that have opened them
divided by a normalization factor I derived from a Harvard Business Review
article I skimmed.

Our readiness score is 74 out of 100.

Seventy-four means we are "Advanced - Approaching Optimization."

Approaching means we haven't arrived.

Optimization means nothing.

The score is on a poster in the lobby of Building 3. It is printed in Helvetica
Neue 120pt. Beneath the score it says "Powered by People, Accelerated by AI." I
wrote that tagline. It won an internal communications award. The award is called
the Catalyst Prize. I also created the Catalyst Prize.

The maturity model has five levels. Level 5 is "AI-Native Organization." Nobody
has reached Level 5. The assessment firm told me this privately. They also told
me the model was designed so that reaching Level 5 requires re-engaging the
assessment firm for "advanced optimization services." The model is a
subscription. The destination is unreachable.

I presented the annual AI strategy to the board last week.

Total spend: $28 million.

Measurable productivity gain: pending.

Pending means we haven't measured it and we've decided that not measuring it is
a strategy rather than an omission.

The board approved a $42 million budget for next year.

They approved it in nine minutes.

Nobody asked what the $28 million accomplished because I showed them the
readiness score and the readiness score went from 58 to 74 and 74 is bigger than
58 and bigger is better and better means we're winning.

I will be promoted to EVP by Q3.

My department has produced one newsletter, one broken chatbot, one meeting tool
that recommends sunsetting customers, one ethics advisor who advises no one, two
roadmaps, three consulting engagements, a maturity score of 1.4, a readiness
score of 74, and a poster in Helvetica Neue.

Goldman Sachs says the impact is basically zero.

My compensation says otherwise.

The graph goes up and to the right. It measures AI investment. Investment means
spending. Spending means commitment. Commitment means we're serious about the
future.

The future is whatever I say it is.

As long as the graph goes up and to the right.