Last month, the Anchorage Assembly passed an ordinance regulating the conditions under which AI data centers can be built in the city. To the casual observer, this probably sounds reasonable. But to read the reaction from the Editors at the Anchorage Daily News, the Assembly is in the process of sealing Anchorage’s economic fate. Grab yourself a coffee or a stiff drink, turn down that AI-generated music, and read on: this one’s a doozy.

If there’s one thing the Editors want you to know, it’s that AI data center regulation = bad.  Anchorage, they write, “is already putting regulations in place on an industry that hasn’t fully landed yet. Guardrails may one day be a good idea, but we shouldn’t be setting up roadblocks before the first truck even arrives.”

First of all, holy mixed metaphors Batman – why are these trucks landing? But more troublingly, the Editors don’t seem to understand that the whole purpose of a guardrail is to prevent trucks from going off the road in the first place. Guardrails are installed by thoughtful engineers who don’t want people to get hurt. If you wait to put necessary guardrails up after the trucks have already careened off the roadway, you’re too late.

But that, bizarrely, is exactly what the Editors are calling for in regards to data center construction: build the data centers first and then sort out appropriate “guardrails” later, after the consequences have become apparent. We “cannot afford to stand around debating hypotheticals while other states move ahead,” they write, as if every aspect of data center construction and operation exists in some hypothetical and unknowable ether.

Conspicuously missing from the Editors’ denunciation of Anchorage’s data center legislation is any explanation of what that legislation actually is, or why passing it was inadvisable. To clear up any confusion, AO 2026-27 included the following:

• A prohibition on data centers in residential areas
• A requirement that data centers enclose machinery to reduce noise
• A requirement that utilities assess power capacity before data centers are approved

The Editors write “Anchorage is proposing unnecessary restrictions that tell the world we’re closed for business.” Got it. So do the Editors believe that data centers should be built in Anchorage without enclosing machinery to reduce noise? Do they want data centers built in Anchorage’s residential neighborhoods? Do they believe that projects should be approved without consideration for railbelt electricity capacity? They don’t say.

When the Editors do address concerns with data centers, their analysis is almost heroically superficial. Consider the following attempt at argument:

“Data centers use a lot of energy — yes, they do. So do most industries that matter.”

AI data centers don’t just “use” energy, they devour in staggering quantities, often driving electricity prices sky-high and causing severe financial hardship and energy grid strain in nearby communities. AI data centers are projected to use 6.7-12% of the nation’s energy capacity by 2028, up from 4.4% just a few years ago. One might as well argue that daycares and fertilizer factories both produce pollution, and therefore should be equivalent from a regulatory perspective.

The Editors repeatedly claim that outside investors are watching Alaska to decide whether to invest here. But they provide no evidence that any companies want to build data centers at scale in Anchorage. Neither do they present a case that our high cost of living, high energy costs, minimal infrastructure, or challenging logistics make Alaska compelling site for data centers at all. Southcentral Alaska is already looking down the barrel of a serious energy crisis. Are the Editors implicitly arguing for the trans-Alaska gas line? Nuclear? Hydro? Again, they don’t say.

What cannot be found in the entirety of the Editors piece is a single coherent, fact-based argument. What can be found, however, are numerous hallmarks of AI writing itself. These include run-on near-meaningless platitudes, generic list-style sentences (“No comprehensive policy, no clear direction, no coordinated strategy.”) and many examples of the telltale “it’s not X, it’s Y” AI writing quirk:

• “…not in theory, but in practice”
• “…it isn’t about hype or fear. It’s about getting smart and doing it quickly.”
• “…growth is not a burden — it’s an economy of scale solution.”

For what it’s worth, this little animal copy/pasted ADN’s editorial into several online AI writing detectors; One analysis reported that the editorial was 100% AI-written. Another reported 84%.

AI writing detectors are far from perfect, but the fact that ADN’s editorial was very likely written in whole or in part by AI does suggest that there is a very different, and arguably unsettling way to see it: this is an AI-written article advocating against extremely reasonable AI regulations.

That sentence should, at minimum, make you ever-so-slightly uncomfortable. AI systems have already demonstrated a keen ability to deceive, manipulate, and even blackmail humans. Major AI companies, desperate to create financially viable business models and escape what’s looking more and more like an AI industry bubble, have outright abandoned their AI safety commitments. And this is the worst the technology will ever be. In the coming years, the ability of AI systems to shape public opinion will only increase.

Is it possible that AI tools have already seduced the Editors at ADN into abandoning their commitment to quality local journalism? The answer to that question is unclear.

What is clear is that this ADN Editorial isn’t a credible critique of data center regulation from a once-Pulitzer-prize-winning newspaper – it’s yet another signpost marking the steady “enshittification” of American journalism. It’s not a persuasive human-centered argument – it’s a sour-smelling buffet of journalistic-ish slop in service of the reckless and predatory AI industry.

It’s not smart and convincing – it’s sad. And a bit disturbing, too.