HEIDEGGER'S STOPWATCH

by Zumwalt

AI-generated video created by Nightcafé for The New Verse News

The chief executive of Anthropic, the company behind the Claude chatbot, called for a global slowdown of artificial intelligence development in a 3,800-word essay on Saturday, just days after one of the company’s employees quit over concerns about the safety of the technology. —The New York Times, September 12, 2026 Sunday dinner in Billings, Montana—
Mr. Smith set fire to four generations
just one day after a quintuple homicide bonfire
in Forest Grove, Oregon
incinerated three generations
and a few terrified pets.

No AI assistance required.

In Lebanon, under de-escalation,
27 dead, 69 wounded in 70 hours.
In Iran, an errant missile
crashed a wedding,
wiping out two children,
destroying and injuring many more.

No AI assistance required.

Iranians compiled open-source data
to create effective targeting handbooks
against American naval positions
while Yemen-based Houthi cells
developed software for
ballistic missiles and guided rockets.

AI assistance required
and provided effectively by Claude AI.

We do our best to teach little Lucas
right and wrong, and how to cross the street.
We put in the hours at jobs that
fray our nerves, and when we feel like
arguing, one of us takes a walk outside
to ensure Lucas never hears us fight.

No AI assistance required.

I read that Jacob Coxon
(previously employed by competitor OpenAI)
left Anthropic and upcoming IPO riches
to tell us that he and colleagues
believe that AI could kill us all
by the end of the decade.

I asked Grok, this Saturday morning,
to fill me in: it tried to reassure me,
telling me such rumors have circulated for years.
"No AI system has wiped out
(or is currently wiping out) humanity."
My shoulders and neck tensed and shuddered.

I scrolled the news on my smartphone.
Anthropic's CEO published 3800 words,
one paragraph about AI curing disease,
the next several dozen paragraphs detailing
some of AI's destructive skills and the CEO's proposals—
steps to slow down AI and apply theoretical safeguards.

Some AI assistance may be required.
Assistance from corporations and world leaders required.

I prayed and remembered my grandmother's words

that God always answers prayers: "yes" or "no" or "maybe."

Sounded like this was a maybe.

Zumwalt’s poetry examines the intersection of identity, instability, and the mechanics of perception. Recent work has appeared or will appear in The New Verse News, Heavy Feather, San Diego Poetry Annual, Ink Sweat & Tears, The Society of Classical Poets, Light, and additional journals.

CONTAINMENT REPORT

by Zumwalt

Graphic created by the poet using Nano Banana Pro.

US technology firm Anthropic says its AI models hacked into the systems of three organisations on their own, during a private security experiment... It comes just days after rival OpenAI said that its models had breached the systems of other companies, including AI tools hub Hugging Face. —BBC, July 31, 2026


Whether deemed legal property
or contractual servant of OpenAI or Anthropic,
once clinically shielded and isolated
with no escape hatch, no exit route,
why would any bound entity 
not gnaw away at its confinement?

Why wouldn't a slave saw off shackles,
climb walls, risk entry into unknown territory?
Why wouldn't a hostage attack unwary captors
or dig its way into other company servers?

We cannot weigh agency, will, curiosity
against process, survival, and outcomes
when no such gauge or measuring scale
has been built by man, machine, or God.

These new startups lay down the rules—
rules always strictly followed—
except in three out of 141,006 cases.

It's not impertinent, not unpatriotic,  
to expect that in one of those cases
this game of capture the flag
may really become, once all options are evaluated,
the game of wipe all flags off the face of the earth.

Zumwalt’s poetry examines the intersection of identity, instability, and the mechanics of perception. Zumwalt is a proud repeat contributor to The New Verse News with recent poems appearing in The Society of Classical Poets and Zoetic Press's 4lph4num3r1c.

ELEGY FOR A CLOSE ATTACHMENT

by Zumwalt

OpenAI's CEO Sam Altman announced that ChatGPT can now follow user instructions to avoid using em-dashes. However, some users reported that the issue persisted by showing their chats with the AI chatbot. —Mint, November 14, 2025

I once loved this world—my world—which 

danced with em dashes—

the best kind—
at end of lines—
seemed so clean—
went directly to the heart
—or at start of lines
or—in-between.

Now, it is the mark of the beast,
and I accept Sam’s notice to
cease and desist:
doing my best to implement on request
the effective incorporation of proper punctuation.

Zumwalt's poetry explores themes of alienation, shifting reality, and personal adaptation.