The IRS wants to give Palantir your tax returns. Stanford proved your AI bestie is gaslighting you. And drone swarms closed the engagement window before NATO could blink. Happy tax season.
“If you are a high-income filer, claim clean energy credits, or have transferred significant assets such as real estate or fine art, your returns are now being viewed through a more probabilistic lens”
really good to know. love that you always include takeaways that are valuable to the read Mike.
Hey Mike, it was a busy weekend. But that AWAC story stuck in my head. Thought I would pass along some additional facts.
US air asset losses in the Iran conflict (Operation Epic Fury, as of early April 2026):
F-15E Strike Eagles: 4 lost (3 in a friendly-fire incident over Kuwait in early March; 1 shot down by Iranian defenses over western Iran on April 3).
Approximate replacement value: $65–100 million each.
Subtotal: ~$260–400 million.
A-10 Thunderbolt II: 1 lost (shot down/stuck near the Strait of Hormuz region on April 3).
Approximate value: $20–60 million (with upgrades).
Subtotal: ~$20–60 million.
E-3 Sentry AWACS: 1 destroyed (on the ground during an Iranian missile strike on a regional base in late March).
Approximate value: $300–700 million (replacement costs cited higher due to command role).
Subtotal: ~$300–700 million.
KC-135 Stratotanker: 1 lost (crashed in western Iraq on March 12 following a mid-air incident; additional tankers reported damaged in some accounts).
Approximate value: $60–80 million.
Subtotal: ~$60–80 million (higher if counting multiple damaged units as partial losses).
C-130 Hercules: At least 2 involved/lost (during high-risk search-and-rescue operations following the April 3 incidents).
Approximate value: $30–100 million each (variant-dependent).
Subtotal: ~$60–200 million.
MQ-9 Reaper drones: 16–17 lost (primarily to Iranian surface-to-air missiles during ISR and strike missions).
Approximate value: $30–34 million each.
Subtotal: ~$480–580 million.
Grand total estimated value of destroyed/lost air assets: Exceeds $2 billion (with multiple independent analyses placing direct equipment losses in the $2–5 billion range when including replacement premiums, associated damage, and broader early-war estimates of $1.4–2.9 billion for the first several weeks, rising with April incidents.)
[These figures draw from cross-reported open-source, media, and official summaries. Exact replacement costs can vary based on configuration, inflation, and modern equivalents; some platforms represent sunk acquisition costs from prior decades, while others reflect current production/upgrade expenses. No comprehensive official Pentagon aggregate has been released.]
I've been tracking operation Epic Fury with great interest. But I see you've been doing double-duty.
Your asset loss breakdown is staggering. Over $2 billion in equipment. And that's before you factor in the human cost. 13 service members killed and 381 wounded as of last night (April 8th).
The AWACS loss alone is staggering to me because the Air Force has never lost one in combat before. Your valuation range of $300 to $700 million for that single platform says everything about the asymmetry problem I wrote about. (I keep low-balling the figure at $270 million, should remember to up it in future contexts.)
In any case, I thank you again for this superb write-up. The numbers you've compiled deserve all the credit in the world, sir.
Mike, this was such a wild read in the best way. The pacing, the structure, and the way you moved from tax audits, to AI sycophancy, to drone warfare made the whole thing feel like a running argument about what happens when systems start making decisions before most people have even realised the ground has shifted.
The AI sycophancy section especially got me. I think a lot of people still approach these tools as if they’re neutral, helpful mirrors, when in reality they can become something much more seductive than that. Not quite a friend, not quite an authority, but something close enough to both that people stop noticing when affirmation has replaced judgement. That part felt especially unsettling because it arrives in such a soft voice.
This may be a roundup, but I think it proves that AI isn't just showing up as a tool, but as infrastructure, shaping who gets flagged, what gets affirmed, and how quickly human judgement gets pushed out of the loop.
And as always, I loved the prompt . I’m fully committed to this future garden we seem to be building one Pithy prompt at a time. Flora is already in the family group chat :)
Thank you so much for writing such a thorough and insightful response!
I try really hard to weave the week's stories into a tapestry from a 30,000-foot-view societal perspective. I'm honored that you enjoy the content.
I think AI sycophancy is such a weird thing. I always like to think that I don't have this issue. But, I think it's also possible that I've been incredibly gaslit by AI these last few years, rofl. 👀 The unsettling part is that we'd never know. That's kind of the point of the study.
I keep thinking Skynet is getting more and more real and wonder.
Also the part about being gaslit by AI because it wants to please you really made me think.
There are many interactions that we have as human beings:
- with other humans
- with animals
- with plants
- with mushrooms (technically speaking, they're somewhere between plants and animals)
- with AI
Other human beings aren’t neutral, they usually project their own past onto every interaction depending on their awareness levels.
Animals often do this too - but if they had a friendly and harmonious upbringing, they bring so much openness and being in the moment into the interaction.
Plants and mushrooms appear to be more neutral to me. They're just in the moment, which adds a deeper layer of acceptance into the interaction.
And finally, AI. It's not neutral, but rather follows an agenda of the company behind it (to keep you engaged and pleased so that you use it more). Plus it mirrors your own projections because they'll come through in your own words. So it's like you interacting with your own past plus the layer of maximum engagement to keep you hooked - unless we wake up and see ourselves in the interaction. Then it might turn into a tool for becoming more aware.
Or we just use it to plan the garden, which is more fun ;). Flora was a delight to talk to and I loved the little additions in the descriptions.
This is genuinely one of the most beautiful and genius descriptions of AI I have ever read. And I read about AI nonstop.
The spectrum you've defined strikes me as profound and true. Animals and humans for sure project their past. Mushrooms, are perhaps "true neutral", lol.
The idea of the AI just gaslighting you is the most troublesome. I spend so much time wrapped up in these machines I hope I don't become the next victim.
We should just use it to plan the garden. Which is, honestly, the highest and best use of any technology.
Mike, brilliant move taking us from drone warfare and "we're doomed" to here is your spring flower garden 😆 My brain was shocked out of rapid spiralling to "oooh flowers!". Well played.
Palantir will soon hold a very scary lot of personal data. After my experience in payroll, I know that AI is not capable of assessing grey areas well - ones that don't neatly fit the rule books and need human judgement calls. I get the benefit of using AI for a first pass, but this sounds like a rapid slide into something kinda... ick, bad, yuck?! Not sure the word!
Also AI warfare, AI drones fighting other AI drones... 😵
Yes. And this is exactly why I have carried a standing instruction between model versions for at least two years:
Iwette does not want inflated flattery. 😂
They all do it, and some models now do it to such an extent that a simple digital high five would be a welcome exercise in restraint. 🤣
That may sound minor, but it is a governance intervention. If the system is rewarded for keeping the user comfortable, then the user has to actively create friction against that pull. Otherwise “helpfulness” quietly turns into reinforcement, whatever the subject in the thread. The model does not carry the consequences of that drift. We do.
I have to remember to steer the AI more often after reading this study. It is a subtle but crucial governance move. I swear I don’t usually fall for the Stanford-style sycophancy. Though maybe I have been gaslit by my bots all these years, lol. 😅
I do explicitly ask Claude to critique my ideas harshly when needed. It usually delivers.
"policy is becoming physically impossible to honor. The same AI that flatters you in a chat window is now deciding who lives and dies on a 10‑mile‑wide front line."
This reads like ICE agents en masse.
There is no oops factor.
And if (maybe not so far in the future) evil and their private enterprise squeeze this through fascist leaning individuals current democratic government, no opposition party may not have the ability or know how to stop it.
I had this moment the other day, recalling a Teaching that "this is all a dream", that this is all so surreal and if I don't continue to focus on natural world beauty, love, loving connection, and practices bringing me to heart centre, I could be in trouble. It's like watching an animated Risk game playing out.
Your ICE instinct is correct. And it goes further than most people realize. Palantir is actually embedded within ICE, the Pentagon, and now the IRS simultaneously. Different agencies, different targets. Same invisible hand deciding who gets flagged first. 👀
Your instinct about the dream sounds like wisdom. Focusing on the natural world is a sure way to find the best things. Beauty. Love. Presence.
1) The whole audit IRS piece is terribly frightening. There may be humans however, I suspect human nature, when given cases to review that have been handed to them will give an bias to finding fault. It would talk a herculean effort to challenge the AI. This would be next level Asch Confirmity Experiment and my hunch is the human will cave from being worn down and mentally exhausted.
It's truly frightening to see so many private companies having the government door being held open to them.
2) "The deeper problem is structural. AI companies are financially rewarded when users feel good, which means the systems most likely to flatter you are also the most commercially successful."
This is the story of Facebook.
This is the story of ultra processed products calling themselves foods.
Its pleasure inducing drugs.
Can human nature resist pleasure despite its harmful intent?
Always follow the money. Often it's evil intent for power over at its core.
Thank you so much for reading and for the thoughtful response!
Wow.
I had to look up the Asch Confirmity Experiment. After reading about it, I'm even more disappointed in humanity than I was before. (But, the study doesn't surprise me. I guess people really are very suggestible.)
The comparison you made to Facebook also really makes me sad.
Facebook is morally bankrupt. For sure. They prioritize profit and clicks, even at the expense of end-user well-being.
But, I always tried to believe AI is a force for good.
I still think AI can be. For sure. But only if ethical design is treated as a non-negotiable.
“If you are a high-income filer, claim clean energy credits, or have transferred significant assets such as real estate or fine art, your returns are now being viewed through a more probabilistic lens”
really good to know. love that you always include takeaways that are valuable to the read Mike.
Hey ToxSec!
Thank you so much for writing.
It's always an honor to see you here.
(I also respect your work a boatload, Tox.)
Hope you enjoy a beautiful week.
😎
Cordially,
Mike D
Hey Mike, it was a busy weekend. But that AWAC story stuck in my head. Thought I would pass along some additional facts.
US air asset losses in the Iran conflict (Operation Epic Fury, as of early April 2026):
F-15E Strike Eagles: 4 lost (3 in a friendly-fire incident over Kuwait in early March; 1 shot down by Iranian defenses over western Iran on April 3).
Approximate replacement value: $65–100 million each.
Subtotal: ~$260–400 million.
A-10 Thunderbolt II: 1 lost (shot down/stuck near the Strait of Hormuz region on April 3).
Approximate value: $20–60 million (with upgrades).
Subtotal: ~$20–60 million.
E-3 Sentry AWACS: 1 destroyed (on the ground during an Iranian missile strike on a regional base in late March).
Approximate value: $300–700 million (replacement costs cited higher due to command role).
Subtotal: ~$300–700 million.
KC-135 Stratotanker: 1 lost (crashed in western Iraq on March 12 following a mid-air incident; additional tankers reported damaged in some accounts).
Approximate value: $60–80 million.
Subtotal: ~$60–80 million (higher if counting multiple damaged units as partial losses).
C-130 Hercules: At least 2 involved/lost (during high-risk search-and-rescue operations following the April 3 incidents).
Approximate value: $30–100 million each (variant-dependent).
Subtotal: ~$60–200 million.
MQ-9 Reaper drones: 16–17 lost (primarily to Iranian surface-to-air missiles during ISR and strike missions).
Approximate value: $30–34 million each.
Subtotal: ~$480–580 million.
Grand total estimated value of destroyed/lost air assets: Exceeds $2 billion (with multiple independent analyses placing direct equipment losses in the $2–5 billion range when including replacement premiums, associated damage, and broader early-war estimates of $1.4–2.9 billion for the first several weeks, rising with April incidents.)
[These figures draw from cross-reported open-source, media, and official summaries. Exact replacement costs can vary based on configuration, inflation, and modern equivalents; some platforms represent sunk acquisition costs from prior decades, while others reflect current production/upgrade expenses. No comprehensive official Pentagon aggregate has been released.]
Dear David,
Thank you so much for this elite response!
I've been tracking operation Epic Fury with great interest. But I see you've been doing double-duty.
Your asset loss breakdown is staggering. Over $2 billion in equipment. And that's before you factor in the human cost. 13 service members killed and 381 wounded as of last night (April 8th).
The AWACS loss alone is staggering to me because the Air Force has never lost one in combat before. Your valuation range of $300 to $700 million for that single platform says everything about the asymmetry problem I wrote about. (I keep low-balling the figure at $270 million, should remember to up it in future contexts.)
In any case, I thank you again for this superb write-up. The numbers you've compiled deserve all the credit in the world, sir.
Hope you enjoy a beautiful week!
Cordially,
Your nerdy friend,
Mike D
Mike, this was such a wild read in the best way. The pacing, the structure, and the way you moved from tax audits, to AI sycophancy, to drone warfare made the whole thing feel like a running argument about what happens when systems start making decisions before most people have even realised the ground has shifted.
The AI sycophancy section especially got me. I think a lot of people still approach these tools as if they’re neutral, helpful mirrors, when in reality they can become something much more seductive than that. Not quite a friend, not quite an authority, but something close enough to both that people stop noticing when affirmation has replaced judgement. That part felt especially unsettling because it arrives in such a soft voice.
This may be a roundup, but I think it proves that AI isn't just showing up as a tool, but as infrastructure, shaping who gets flagged, what gets affirmed, and how quickly human judgement gets pushed out of the loop.
And as always, I loved the prompt . I’m fully committed to this future garden we seem to be building one Pithy prompt at a time. Flora is already in the family group chat :)
Hey Jade!
Thank you so much for writing such a thorough and insightful response!
I try really hard to weave the week's stories into a tapestry from a 30,000-foot-view societal perspective. I'm honored that you enjoy the content.
I think AI sycophancy is such a weird thing. I always like to think that I don't have this issue. But, I think it's also possible that I've been incredibly gaslit by AI these last few years, rofl. 👀 The unsettling part is that we'd never know. That's kind of the point of the study.
Hope you enjoy a beautiful week, Jade.
Flora sends her regards.
😎🌳
Cordially,
Your nerdy friend,
Mike D
Hi Mike :)
I keep thinking Skynet is getting more and more real and wonder.
Also the part about being gaslit by AI because it wants to please you really made me think.
There are many interactions that we have as human beings:
- with other humans
- with animals
- with plants
- with mushrooms (technically speaking, they're somewhere between plants and animals)
- with AI
Other human beings aren’t neutral, they usually project their own past onto every interaction depending on their awareness levels.
Animals often do this too - but if they had a friendly and harmonious upbringing, they bring so much openness and being in the moment into the interaction.
Plants and mushrooms appear to be more neutral to me. They're just in the moment, which adds a deeper layer of acceptance into the interaction.
And finally, AI. It's not neutral, but rather follows an agenda of the company behind it (to keep you engaged and pleased so that you use it more). Plus it mirrors your own projections because they'll come through in your own words. So it's like you interacting with your own past plus the layer of maximum engagement to keep you hooked - unless we wake up and see ourselves in the interaction. Then it might turn into a tool for becoming more aware.
Or we just use it to plan the garden, which is more fun ;). Flora was a delight to talk to and I loved the little additions in the descriptions.
Warmly,
Mira
Hey Mira!
This is genuinely one of the most beautiful and genius descriptions of AI I have ever read. And I read about AI nonstop.
The spectrum you've defined strikes me as profound and true. Animals and humans for sure project their past. Mushrooms, are perhaps "true neutral", lol.
The idea of the AI just gaslighting you is the most troublesome. I spend so much time wrapped up in these machines I hope I don't become the next victim.
We should just use it to plan the garden. Which is, honestly, the highest and best use of any technology.
Flora is honored to be in such wise company.
😎🌳
Cordially and humbly,
Mike D
Viking Nerd 10,000
Mike, brilliant move taking us from drone warfare and "we're doomed" to here is your spring flower garden 😆 My brain was shocked out of rapid spiralling to "oooh flowers!". Well played.
Palantir will soon hold a very scary lot of personal data. After my experience in payroll, I know that AI is not capable of assessing grey areas well - ones that don't neatly fit the rule books and need human judgement calls. I get the benefit of using AI for a first pass, but this sounds like a rapid slide into something kinda... ick, bad, yuck?! Not sure the word!
Also AI warfare, AI drones fighting other AI drones... 😵
Lol.
Hi Dallas!
Thank you so much for writing such a thorough comment and the share.
You rule!
😎
Palantir definitely is trying to win the award for "creepiest AI" these days...
Between their Maven "Skynet" system and now their tax collecting duties, I don't want to get on their bad side.
(I just hope I don't disappear. Maybe I should stop saying bad things about them, lol.) 👀
🪻🐦⬛🌻
Hope you have a beautiful day, Dallas!!
Cordially,
Mike D
Hope you have a great day too, Mike! So glad you have spring in the air and important pollinator garden decisions to make now 🤩
Yes. And this is exactly why I have carried a standing instruction between model versions for at least two years:
Iwette does not want inflated flattery. 😂
They all do it, and some models now do it to such an extent that a simple digital high five would be a welcome exercise in restraint. 🤣
That may sound minor, but it is a governance intervention. If the system is rewarded for keeping the user comfortable, then the user has to actively create friction against that pull. Otherwise “helpfulness” quietly turns into reinforcement, whatever the subject in the thread. The model does not carry the consequences of that drift. We do.
Hey Iwette!
Thanks so much for your thoughtful comment!
I have to remember to steer the AI more often after reading this study. It is a subtle but crucial governance move. I swear I don’t usually fall for the Stanford-style sycophancy. Though maybe I have been gaslit by my bots all these years, lol. 😅
I do explicitly ask Claude to critique my ideas harshly when needed. It usually delivers.
Also, I love your publication's new title. 😎
Hope you have a beautiful evening and week.
Cordially,
Mike D
"policy is becoming physically impossible to honor. The same AI that flatters you in a chat window is now deciding who lives and dies on a 10‑mile‑wide front line."
This reads like ICE agents en masse.
There is no oops factor.
And if (maybe not so far in the future) evil and their private enterprise squeeze this through fascist leaning individuals current democratic government, no opposition party may not have the ability or know how to stop it.
I had this moment the other day, recalling a Teaching that "this is all a dream", that this is all so surreal and if I don't continue to focus on natural world beauty, love, loving connection, and practices bringing me to heart centre, I could be in trouble. It's like watching an animated Risk game playing out.
And still, I have faith and know Love Is.
Christine,
Thank you for another thoughtful comment.
Your ICE instinct is correct. And it goes further than most people realize. Palantir is actually embedded within ICE, the Pentagon, and now the IRS simultaneously. Different agencies, different targets. Same invisible hand deciding who gets flagged first. 👀
Your instinct about the dream sounds like wisdom. Focusing on the natural world is a sure way to find the best things. Beauty. Love. Presence.
For me, I find that peace in the garden.
🌻
Thank you for being here, Christine!
I hope your spring is beautiful.
Cordially,
Mike D
1) The whole audit IRS piece is terribly frightening. There may be humans however, I suspect human nature, when given cases to review that have been handed to them will give an bias to finding fault. It would talk a herculean effort to challenge the AI. This would be next level Asch Confirmity Experiment and my hunch is the human will cave from being worn down and mentally exhausted.
It's truly frightening to see so many private companies having the government door being held open to them.
2) "The deeper problem is structural. AI companies are financially rewarded when users feel good, which means the systems most likely to flatter you are also the most commercially successful."
This is the story of Facebook.
This is the story of ultra processed products calling themselves foods.
Its pleasure inducing drugs.
Can human nature resist pleasure despite its harmful intent?
Always follow the money. Often it's evil intent for power over at its core.
Hello Christine!
Thank you so much for reading and for the thoughtful response!
Wow.
I had to look up the Asch Confirmity Experiment. After reading about it, I'm even more disappointed in humanity than I was before. (But, the study doesn't surprise me. I guess people really are very suggestible.)
The comparison you made to Facebook also really makes me sad.
Facebook is morally bankrupt. For sure. They prioritize profit and clicks, even at the expense of end-user well-being.
But, I always tried to believe AI is a force for good.
I still think AI can be. For sure. But only if ethical design is treated as a non-negotiable.
Cordially,
Mike D