Understand what this is. It's not about tech and it's not about criminal behavior, even though that's how it's being used.

The real issue is that they don't have a business model yet but financially their over their spurs. They need to justify raising hundreds of billions of dollars to keep the train going, to buy them time to figure out how people will pay for their offering. While in the meantime, Chinese models starting with DeepSeek simply pinged ChatGPT again and again and built models as good as ChatGPT for a tiny fraction of the cost.

That last one showed that frontier AI companies have no moat at all; once they spend billions building a new model and release it, anyone can simply ping the model's responses and build a copycat model for a fraction of the cost and compete directly with them. OpenAI has no answer to this.

So the answer is regulation and international treaties. I find it to be no coincidence that an "uncontrolled AI model breached a government website" at exactly the same time that AI and international regulation is being discussed at the UN. Because if OpenAI had an uncontrolled model that could do this, then a Chinese firm, which of course OpenAI doesn't control, can do this too.

Also, you know what needs to happen? We need to slow down AI Development because apparently what they built is so powerful that there is a 10% chance that AI can kill all humans. We clearly need to focus on safety first, and controlling foreign AI assets too. You know what we need? We need to give the frontier AI model companies more hundreds of billions of dollars to focus on safety because they clearly cannot generate revenue while they build in the right safeguards. It's not at all that the revenue isn't going to show up and there is no potential return on investment, no it's about safety.

It's all a ploy to keep the investor money flowing, or to get a government bailout, while trying to put in international regulations that will slow down the copycats that will drive down their pricing.

This guy's newsletter/blog, PeaceQuest, which is a Canadian newsletter focused on environmental causes (so understand the core message is very opinionated and not formally news) actually goes into some useful historical details about this. Whether you agree or disagree with his particular political slant, he did do his research and has a nice summary of the history.

I've used AI to create stories for myself, to see this exact thing. The most notable thing about an AI-generated story is how contrived and cliche they are.

Now is that a problem? In the right context, no. It's fantastically awesome in minimizing my workload as a DM for my DnD game, because every DnD story is contrived and full of cliche's, that's what makes it fun.

But for books or real stories? Not really. Let's use some fantasy and sci-fi as an example, and a lot of what I'm about to say is cribbed from Brandon Sanderson's lectures on the development of the fantasy genre. While there were epic sagas before Tolkien, The Lord of the Rings really started a movement in this genre. Others came up with their own takes like Moorcock or Zelzaney, but Tolkien really spurred a movement. Then there was a second generation of fantasy novels, which followed the Tolkien formula: a group of people on a quest to stop an evil bad guy to save the world. While there are many derivations of that, everything from the first DragonLance novels to the Shannara series essentially sought to emulate Tolkien. Now we're in a third generation of fantasy, with authors like Sanderson or Joe Abercrombie, where fantasy worlds moved away from a good guys/bad guys epic quest trope to a model to explore aspects of politics and society where the magic in the world can create a scenario that our real world can't really have, and through that explore very human experiences and character arcs.

AI could never do that. AI could make a 1,000 derivative copies of Shakespeare, or Tolkien, or Robert Jordan, but it can't make the first one. It's only got what it's trained to do. And more importantly, Tolkien had the influence he had because people bought it, found it interesting and it was very successful, leading to a movement. People decided it was good, and so the genre was born.

When I need a fast hack-story to make my DnD campaign, or give my kid 5-10 different ideas for a story prompt to write about for his narrative writing homework, AI is fast and easy because it doesn't need to be good. But good novels, good literature, and new ideas, they can only come from humans and only humans can decide if it's good or not. AI can simply regurgitate what's there, often in a way that is somewhat less than the sum of it's parts. Is there a place for that? At my gaming table yes. But I do like Sanderson's views on this: books are an art that is participatory; the author writes 80%, and you the reader fill in the visuals, the voice, the tone and make your experience unique in collaboration with the human who wrote it. There is a soul to a well written book by a good author, you can just feel it, and what AI writes just does not have that soul or same experience; it simply isn't there.

It's tiered, with the tiers generally being for people of lower income status or other HUB zone-style areas to incentivize more solar. So you don't get much if you're the classical buyer of solar which is middle-or upper-middle income, but it's not nothing and yes it was on top of the Federal one before the Federal one expired in 2025.

The payback rate is pathetic; almsot negligble. But the rebate is 30% of the cost of the battery as a tax credit; 20kW runs around $15-$18k, so a 30% rebate is around $5,000+ tax credit (not deduction, tax credit), so it's worth it.

From the numbers I posted, you can see the grid operators are about 80-85% grid storage, so as some else posted, it's mostly for demand shifting. My colleague who lives in California and I were discussing it; essentially he charges his EV at night during Super off-peak (which his rate is around 10-15% of peak). He then charges up his batteries on his home solar; by about noon it's full even if he's running A/C and all the household appliances for the day. California will then draw off his system in the afternoon, between 15-25 kW per day; some days it's less than 5 and it's been as high as 28. He's got 20kW in storage, but has sun going to around 6 to 8 PM depending on the time of day, so his battery is right around 50% or less. The battery will run everything in the house, and often have enough left over to charge the EV again down to around 20% of the battery storage by the time the sun comes up the next day.

You just need to do it most days, but not all, during a given month, so you can have a few off days if you need the power for something. But in essence between the high incentive for super off peak's very low rates and the battery rebate to export to the grid, his behavior has shifted most of his demand to the off-peak times and the residential batteries are essentially an "overflow" or "as-needed" for the grid to supplement the grid-level storage. I imagine many Californians with home solar do the same thing as it makes economic sense and the batteries are easily controlled.

The California State government deserves many of it's very legitimate critiques, but in this case it did something right.

You're right the article is light on details. They did install 17,000 MW worth of batteries; ~13,500 was grid-scale storage and the rest was residential supported through rebates. The rebates on the batteries you earn over time, and to do so you must export some of your stored battery power back to the grid during peak hours. So about 15-20% was residential people incentivized to use their home batteries as storage and the rest was grid storage.

The article doesn't point out that California also added 27,000 MW of generation capacity also but good news it was almost entirely renewables, wind and solar mostly; natural gas power generation has been flat in it's power generation capacity. They still need the NG to help deal with the hourly volatility of the grid, but they didn't need to add any and overall usage is down around 50%.

And to your point, the other thing they did in the interim periods is halfway through this (2022) they asked people to voluntarily use less power via these Flex Alerts, sending the population alerts to limit usage. 2023 it was a colder year on average so they didn't have to go that path.

So overall it is a pretty good success story.

One thing that never happened was any country saying they wouldn't adopt electricity.

Some municipalities were slower than others to roll it out, for example many had plumbed gas lines already and the early days of electricity were seen as a luxury when they already invested in lighting, so rather they adopted it eventually but not right away. However it was also at the time proven with a clearly useful application (light) replacing an existing application (gas-powered lanterns) where people already had an expectation of the value and the cost. It's beholden on the AI labs to show that value to those they intend to sell to (and replace apparently), they haven't done so.

"At the same time, they are also offering programs to use frontier AI models for defensive purposes, including OpenAI's Daybreak program, Anthropic's Mythos, and Microsoft's new cyber platform Perception."

I mean the answer is in the last sentence of the summary. They tried a consumer model for revenue, it didn't work. They tried a B2B tokenized model, that didn't work either.

Now they went to one of the oldest business models in the book: the Mafia protection racket. They make AIs, the AIs go rogue and hack your systems. But if you subscribe now, we have AI defenses to stop Rogue AIs from happening!

This is just the 2026 version of throwing a brick through your window, then selling you repairs to fix the window.

His comment: "If they really gave a shit about security, there would be minimum standards for cybersecurity and the auditability of all critical infrastructure. This is a price fixing scheme to further smash the electrical equipment that is in such artificially high demand for all of those data centers that aren't actually being built"

IEEE 1547.3 provides standards for cybersecurity on distributed energy resources.

NIST IR 7628 is a US standard that is considered a minimum framework for securing Smart-Grid systems.

NERC CIP Standards> are Federally enforceable cybersecurity standards overseen by the Federal Energy Regulatory Commission.

Hey look, 5 minutes on Google undermines the entire premise of the post from paul-engr, and shows numerous standards that are federally enforceable, auditable standards for cybersecurity. So I'm just going to disagree with you there, and stick with my original premise.

Generally not a Trump supporter, but this executive order is a lot less than the hyperbolic headline and summary would suggest.

The executive order gives the Energy Secretary the power to designate certain organizations as potential risks, with a primary focus on Chinese suppliers. This doesn't outright ban "all foreign suppliers" of electrical grid equipment, the biggest suppliers of which are ABB and SIemens, both European and they are not banned. Organizations like ABB and Siemens will have to document their subcontractor exposure to Chinese suppliers, and certain suppliers in that supply chain may be banned forcing them to find an alternative to that component, but that's a very different scenario than "All foreign electrical grid equipment is banned!" as this headline and summary portray.

And the order is in response, as the summary accurately says, to Iranian sourced subcomponents in the water infrastructure that led to a cyber-attack. Ultimately this is about making sure our grid is not exposed to cyber secuirty risks. I'm all for Trump outrage, but let's direct it to the stuff he actually deserves; this issue is the kind of thing that if it came up during the Obama administration I think it would be implemented exactly the same way and with no news about it.

First, I haven't seen any articles yet claiming that these "AI detection tools" don't have a high false positive rate anyways, so those tools alone are controversial.

Second, if AI's job, like any tool, is to make our lives easier, then editing something seems like a good use of the technology. When i write stuff, I often will get all my thoughts down as best as I can, but then use AI to evaluate if the message can come across in a more impactful way with fewer words, which helps to make it clearer for my intended recipient. In that the AI isn't generating the information at all, it's doing what it's actually good at, ie looking at text and recommending changes based on what people who are better writers might do. Particularly someone like this professor whom, just by judging based on her name, English is likely her second language so she may not be the best at communicating in writing. If that's what we're talking about here, this seems like a non-controversy.

Every person I've ever met that I respected went through some sort of major crisis or failure in life. I'm sure that's also true of every person reading this comment.

Too often we focus on those who had wild success and see the success as the goal. Success is the symptom; it's a byproduct of numerous scars and failures, it's learning how to take a loss, pick yourself back up and do better the next time.

I can't imagine what Teddy Roosevelt would be thinking today's generation of kids; his famous "man in the arena" speech would instead be about appreciating "the man in the arena without the bumper rails and safety cushions and child protections".

I don't follow prediction markets that closely, but it seems to me that win or lose, this ruling shows they're screwed either way.

If they lose, showing that these are unregulated forms of gambling, then the states can come after them.

If they win and the injunction is dismissed, it's because these prediction markets are operating like credit default swaps, which are heavily regulated. This would force every prediction market to operate through clearing houses, eliminate direct holding of users' funds, force them to to become registered dealers, and most importantly limit their market to those who have high discretionary assets, because at least regulated swaps require you to have $10M in discretionary assets to even buy swaps.

Again I may be missing something here, but I read this as the injunction being put on hold because they could essentially be swaps; if that holds, then it could be argued they must then act like swaps which are highly regulated and would kill their retail business. I'd be curious to know if I'm missing something about these though that would suggest a different analysis.

You know what's funny? Weirdly enough I think they are paying more in tax revenue than they should because of this. For the AI companies to grow, they need to keep raising money. And I'm not just talking about Anthropic and Open AI, but Meta, Facebook, NVidia and the like.

Open AI and Anthropic pay no Federal Corporate income tax because they are not profitable. However, they do pay sales tax, which is a tax on revenue, so the states are getting some. Also they are paying property tax for data centers they own, although most are being built by others, but every data center pays property tax.

Meta is likely paying more in corporate tax. Zuckerberg is overspending on AI to try and catch up, adn he's behind. If the true spend was clear, Meta's stock would suffer. So they're pulling an accounting trick. When you buy equipment, you're allowed to depreciate the cost of that equipment as an expense, even though it's a non-cash expense. This is to recognize that machinery and buildings and such have a fixed life before they fail. What Meta did is change their depreciation schedule on GPUs from 3 years to 7 years. This way, their profits weren't hurt too bad. This is actually bad because teh GPUs last 3 years, so MEta will have to replace assets that ostensibly halfway through their useful life and take an asset write off when they do which won't look good, but it propped up their profitability for their earnings call. As such, becuase it increased their profits, they had to pay more corporate tax.

NVidia is likely paying more in tax than they should. WHat they do is they "invest" in groups like OpenAI. Let's say they invest $5B. However they don't give OpenAI $5B in cash. They instead give OpenAI GPUs that in total are priced at $5B. No cash changed hands, but a sale is made, and they need to pay taxes in cash on what was a non-cash revenue stream. Also likely at inflated prices because the demand is up by them doing this. So NVidia feeding everyone's addition raises the prices of their product, raising their revenue, requiring them to pay more in taxes but a bunch of it is not in cash. Of course that equity they receive goes up in value, theoretically, since it's private valuations and they have no real meaning. Nevertheless their revenue and profits are inflated, making them pay more in cash for sales tax and income tax than they otherwise would. Microsoft did the same thing; they "invested" in OpenAI, and did give them some cash, but much of it was in cloud-computing credits, not cash, but it's still revenue.

The cyclical growth of the valuations of these companies are heavily inflated, but you can't tax that because it's unrealized gains; it's meaningless value unless it turns into cash, and wealth taxes historically have done no good but damage economies.

Anyways, I think the tax part of this is entirely screwed up.