I Like The Cryptos

i just like googlin stuff

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@kellyclarkson Are you on drugs you fucking retard? You didn’t even say that they directly told you that you couldn’t afford it. “Gave me a look”. So you’re just a sensitive assumptive fucking moron with an ego. Good job dipshit I will definitely check them out and buy things 👍🏼

[I Like The Cryptos @LikeCryptos]

@spencerpratt You’ll see more of PauseAI. Close ties to Coxon.

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@bneiluj My bad this is where I found Hubinger

[I Like The Cryptos @LikeCryptos]

UPDATE: The network around Jacob Coxon goes back much further than Anthropic.

I've been mapping the people who knew Coxon, publicly validated his story, and connect into the AI-safety advocacy ecosystem.

The earliest connection I've found goes back a decade.

In 2016, the UK sent six students to the International Mathematical Olympiad.

Three of them were:

Jacob Coxon
Joe Benton
Neel Nanda
All three made the six-person UK team AGAIN in 2017.
This isn't “they went to similar schools.”

Coxon, Benton and Nanda were teammates representing the UK at the highest level of international competitive mathematics for two consecutive years.

Ten years later, all three converge in the frontier-AI/safety world.

Coxon → OpenAI → Anthropic.

Benton → UK Frontier AI Taskforce → Anthropic → METR.

Nanda → Anthropic → Google DeepMind, where he leads mechanistic interpretability research.

Now follow Nanda.

Nanda directly connects that old Coxon/Benton network to Joseph Miller.

Nanda and Miller are both authors on the 2025 “Open Problems in Mechanistic Interpretability” paper.

Miller also came through MATS, while Nanda is one of MATS's most prolific mentors and says he's worked with roughly 60 scholars.

Miller is now Director of PauseAI UK.

So there is a documented path:

COXON ↔ NANDA ↔ MILLER ↔ PAUSEAI UK

And Miller is the guy who, after Coxon's resignation, told AI-lab employees:

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He offered to help execute such strategies and subsequently added:

“It seems that Jacob Coxon did have a media strategy and it paid off.”

But there's another side of the network.

COXON ↔ BENTON ↔ ANTHROPIC ALIGNMENT

Benton publicly backed Coxon's characterization after Coxon resigned.

Evan Hubinger, one of Anthropic's senior alignment researchers, also immediately backed Coxon publicly.

And now add Ethan Perez.

Perez says he and other Anthropic colleagues had been trying to recruit Coxon for roughly TWO YEARS.

More importantly, Perez says he personally spoke to Coxon a couple of weeks before Coxon resigned.

According to Perez, Coxon was already deeply concerned about Anthropic's pace of progress. Perez tried to convince him to stay and join his team.

Coxon reportedly told him that more than simply doing safety research was necessary.

Then Coxon resigns.

His media strategy launches.

And Perez, Hubinger and Benton all publicly validate important parts of his story.

That gives us another network:

BENTON ↔ COXON ↔ PEREZ
↕ ↕
ANTHROPIC ALIGNMENT ↔ HUBINGER

And these aren't isolated from the first network.

Nanda is deeply involved in MATS and the broader alignment community.
Hubinger was also part of the early MATS ecosystem.
Miller was a MATS Fellow.

Nanda and Miller share a research relationship.

Nanda, Hubinger and Perez also appear in the broader AI-safety grantmaking/regranting ecosystem.

So zoom all the way out:

UK IMO 2016/17
↓
COXON ↔ BENTON ↔ NANDA
↓ ↓
ANTHROPIC MATS / ALIGNMENT
↓ ↓
PEREZ ↔ HUBINGER MILLER
↓
PAUSEAI

Then look at what happens in September 2026.

Coxon executes what Miller himself describes as a successful “media strategy.”

Hubinger validates Coxon's central claim.

Perez reveals he'd been discussing Coxon's concerns with him BEFORE the resignation.
Benton backs Coxon's description.
Miller points to the resulting spike in PauseAI UK sign-ups and immediately proposes scaling the mechanism:
“If one previously-unknown Anthropic employee had this much impact, imagine the followup article: ‘4 Anthropic employees and 3 OpenAI employees join Jacob Coxon in quitting.’ Think it through.”

The question is no longer whether these people inhabit the same general AI-safety universe.

They demonstrably do.
Some of these relationships go back TEN YEARS.
The question now is much narrower:
Who knew about Coxon's planned media campaign before it launched, and who helped him execute it?

Sources/receipts below.
https://t.co/LGZLRlK8He
https://t.co/Le479mwehd
https://t.co/5pSW0q0W17

[I Like The Cryptos @LikeCryptos]

@bneiluj Hubinger and Coxon have been in cahoots for a while. Can’t trust the ones that immediately hopped on his posts to give him credibility. It’s legitimately been in plan for a decade. https://t.co/Z8y0a8Ui7z

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@MJTruthUltra I have another post with an update on it but yeah this is a setup. 10 years in the making.

https://t.co/Z8y0a8Ui7z

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@EllieAsksWhy @petrowashere Miller and Coxon were introduced through Nanda. I think I responded to your other comment with my latest update that breaks it down.

[I Like The Cryptos @LikeCryptos]

UPDATE: The network around Jacob Coxon goes back much further than Anthropic.

I've been mapping the people who knew Coxon, publicly validated his story, and connect into the AI-safety advocacy ecosystem.

The earliest connection I've found goes back a decade.

In 2016, the UK sent six students to the International Mathematical Olympiad.

Three of them were:

Jacob Coxon
Joe Benton
Neel Nanda
All three made the six-person UK team AGAIN in 2017.
This isn't “they went to similar schools.”

Coxon, Benton and Nanda were teammates representing the UK at the highest level of international competitive mathematics for two consecutive years.

Ten years later, all three converge in the frontier-AI/safety world.

Coxon → OpenAI → Anthropic.

Benton → UK Frontier AI Taskforce → Anthropic → METR.

Nanda → Anthropic → Google DeepMind, where he leads mechanistic interpretability research.

Now follow Nanda.

Nanda directly connects that old Coxon/Benton network to Joseph Miller.

Nanda and Miller are both authors on the 2025 “Open Problems in Mechanistic Interpretability” paper.

Miller also came through MATS, while Nanda is one of MATS's most prolific mentors and says he's worked with roughly 60 scholars.

Miller is now Director of PauseAI UK.

So there is a documented path:

COXON ↔ NANDA ↔ MILLER ↔ PAUSEAI UK

And Miller is the guy who, after Coxon's resignation, told AI-lab employees:

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He offered to help execute such strategies and subsequently added:

“It seems that Jacob Coxon did have a media strategy and it paid off.”

But there's another side of the network.

COXON ↔ BENTON ↔ ANTHROPIC ALIGNMENT

Benton publicly backed Coxon's characterization after Coxon resigned.

Evan Hubinger, one of Anthropic's senior alignment researchers, also immediately backed Coxon publicly.

And now add Ethan Perez.

Perez says he and other Anthropic colleagues had been trying to recruit Coxon for roughly TWO YEARS.

More importantly, Perez says he personally spoke to Coxon a couple of weeks before Coxon resigned.

According to Perez, Coxon was already deeply concerned about Anthropic's pace of progress. Perez tried to convince him to stay and join his team.

Coxon reportedly told him that more than simply doing safety research was necessary.

Then Coxon resigns.

His media strategy launches.

And Perez, Hubinger and Benton all publicly validate important parts of his story.

That gives us another network:

BENTON ↔ COXON ↔ PEREZ
↕ ↕
ANTHROPIC ALIGNMENT ↔ HUBINGER

And these aren't isolated from the first network.

Nanda is deeply involved in MATS and the broader alignment community.
Hubinger was also part of the early MATS ecosystem.
Miller was a MATS Fellow.

Nanda and Miller share a research relationship.

Nanda, Hubinger and Perez also appear in the broader AI-safety grantmaking/regranting ecosystem.

So zoom all the way out:

UK IMO 2016/17
↓
COXON ↔ BENTON ↔ NANDA
↓ ↓
ANTHROPIC MATS / ALIGNMENT
↓ ↓
PEREZ ↔ HUBINGER MILLER
↓
PAUSEAI

Then look at what happens in September 2026.

Coxon executes what Miller himself describes as a successful “media strategy.”

Hubinger validates Coxon's central claim.

Perez reveals he'd been discussing Coxon's concerns with him BEFORE the resignation.
Benton backs Coxon's description.
Miller points to the resulting spike in PauseAI UK sign-ups and immediately proposes scaling the mechanism:
“If one previously-unknown Anthropic employee had this much impact, imagine the followup article: ‘4 Anthropic employees and 3 OpenAI employees join Jacob Coxon in quitting.’ Think it through.”

The question is no longer whether these people inhabit the same general AI-safety universe.

They demonstrably do.
Some of these relationships go back TEN YEARS.
The question now is much narrower:
Who knew about Coxon's planned media campaign before it launched, and who helped him execute it?

Sources/receipts below.
https://t.co/LGZLRlK8He
https://t.co/Le479mwehd
https://t.co/5pSW0q0W17

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@EllieAsksWhy @officialrari @ParkerThayer They’re all in cahoots.

[I Like The Cryptos @LikeCryptos]

UPDATE: The network around Jacob Coxon goes back much further than Anthropic.

I've been mapping the people who knew Coxon, publicly validated his story, and connect into the AI-safety advocacy ecosystem.

The earliest connection I've found goes back a decade.

In 2016, the UK sent six students to the International Mathematical Olympiad.

Three of them were:

Jacob Coxon
Joe Benton
Neel Nanda
All three made the six-person UK team AGAIN in 2017.
This isn't “they went to similar schools.”

Coxon, Benton and Nanda were teammates representing the UK at the highest level of international competitive mathematics for two consecutive years.

Ten years later, all three converge in the frontier-AI/safety world.

Coxon → OpenAI → Anthropic.

Benton → UK Frontier AI Taskforce → Anthropic → METR.

Nanda → Anthropic → Google DeepMind, where he leads mechanistic interpretability research.

Now follow Nanda.

Nanda directly connects that old Coxon/Benton network to Joseph Miller.

Nanda and Miller are both authors on the 2025 “Open Problems in Mechanistic Interpretability” paper.

Miller also came through MATS, while Nanda is one of MATS's most prolific mentors and says he's worked with roughly 60 scholars.

Miller is now Director of PauseAI UK.

So there is a documented path:

COXON ↔ NANDA ↔ MILLER ↔ PAUSEAI UK

And Miller is the guy who, after Coxon's resignation, told AI-lab employees:

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He offered to help execute such strategies and subsequently added:

“It seems that Jacob Coxon did have a media strategy and it paid off.”

But there's another side of the network.

COXON ↔ BENTON ↔ ANTHROPIC ALIGNMENT

Benton publicly backed Coxon's characterization after Coxon resigned.

Evan Hubinger, one of Anthropic's senior alignment researchers, also immediately backed Coxon publicly.

And now add Ethan Perez.

Perez says he and other Anthropic colleagues had been trying to recruit Coxon for roughly TWO YEARS.

More importantly, Perez says he personally spoke to Coxon a couple of weeks before Coxon resigned.

According to Perez, Coxon was already deeply concerned about Anthropic's pace of progress. Perez tried to convince him to stay and join his team.

Coxon reportedly told him that more than simply doing safety research was necessary.

Then Coxon resigns.

His media strategy launches.

And Perez, Hubinger and Benton all publicly validate important parts of his story.

That gives us another network:

BENTON ↔ COXON ↔ PEREZ
↕ ↕
ANTHROPIC ALIGNMENT ↔ HUBINGER

And these aren't isolated from the first network.

Nanda is deeply involved in MATS and the broader alignment community.
Hubinger was also part of the early MATS ecosystem.
Miller was a MATS Fellow.

Nanda and Miller share a research relationship.

Nanda, Hubinger and Perez also appear in the broader AI-safety grantmaking/regranting ecosystem.

So zoom all the way out:

UK IMO 2016/17
↓
COXON ↔ BENTON ↔ NANDA
↓ ↓
ANTHROPIC MATS / ALIGNMENT
↓ ↓
PEREZ ↔ HUBINGER MILLER
↓
PAUSEAI

Then look at what happens in September 2026.

Coxon executes what Miller himself describes as a successful “media strategy.”

Hubinger validates Coxon's central claim.

Perez reveals he'd been discussing Coxon's concerns with him BEFORE the resignation.
Benton backs Coxon's description.
Miller points to the resulting spike in PauseAI UK sign-ups and immediately proposes scaling the mechanism:
“If one previously-unknown Anthropic employee had this much impact, imagine the followup article: ‘4 Anthropic employees and 3 OpenAI employees join Jacob Coxon in quitting.’ Think it through.”

The question is no longer whether these people inhabit the same general AI-safety universe.

They demonstrably do.
Some of these relationships go back TEN YEARS.
The question now is much narrower:
Who knew about Coxon's planned media campaign before it launched, and who helped him execute it?

Sources/receipts below.
https://t.co/LGZLRlK8He
https://t.co/Le479mwehd
https://t.co/5pSW0q0W17

[I Like The Cryptos @LikeCryptos]

@petrowashere 100%. I put some more posts up about my findings on Coxon and where I think this is coming from if you’re interested in that rabbit hole.

[I Like The Cryptos @LikeCryptos]

@ParkerThayer you seem interested in this

[I Like The Cryptos @LikeCryptos]

UPDATE: The network around Jacob Coxon goes back much further than Anthropic.

I've been mapping the people who knew Coxon, publicly validated his story, and connect into the AI-safety advocacy ecosystem.

The earliest connection I've found goes back a decade.

In 2016, the UK sent six students to the International Mathematical Olympiad.

Three of them were:

Jacob Coxon
Joe Benton
Neel Nanda
All three made the six-person UK team AGAIN in 2017.
This isn't “they went to similar schools.”

Coxon, Benton and Nanda were teammates representing the UK at the highest level of international competitive mathematics for two consecutive years.

Ten years later, all three converge in the frontier-AI/safety world.

Coxon → OpenAI → Anthropic.

Benton → UK Frontier AI Taskforce → Anthropic → METR.

Nanda → Anthropic → Google DeepMind, where he leads mechanistic interpretability research.

Now follow Nanda.

Nanda directly connects that old Coxon/Benton network to Joseph Miller.

Nanda and Miller are both authors on the 2025 “Open Problems in Mechanistic Interpretability” paper.

Miller also came through MATS, while Nanda is one of MATS's most prolific mentors and says he's worked with roughly 60 scholars.

Miller is now Director of PauseAI UK.

So there is a documented path:

COXON ↔ NANDA ↔ MILLER ↔ PAUSEAI UK

And Miller is the guy who, after Coxon's resignation, told AI-lab employees:

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He offered to help execute such strategies and subsequently added:

“It seems that Jacob Coxon did have a media strategy and it paid off.”

But there's another side of the network.

COXON ↔ BENTON ↔ ANTHROPIC ALIGNMENT

Benton publicly backed Coxon's characterization after Coxon resigned.

Evan Hubinger, one of Anthropic's senior alignment researchers, also immediately backed Coxon publicly.

And now add Ethan Perez.

Perez says he and other Anthropic colleagues had been trying to recruit Coxon for roughly TWO YEARS.

More importantly, Perez says he personally spoke to Coxon a couple of weeks before Coxon resigned.

According to Perez, Coxon was already deeply concerned about Anthropic's pace of progress. Perez tried to convince him to stay and join his team.

Coxon reportedly told him that more than simply doing safety research was necessary.

Then Coxon resigns.

His media strategy launches.

And Perez, Hubinger and Benton all publicly validate important parts of his story.

That gives us another network:

BENTON ↔ COXON ↔ PEREZ
↕ ↕
ANTHROPIC ALIGNMENT ↔ HUBINGER

And these aren't isolated from the first network.

Nanda is deeply involved in MATS and the broader alignment community.
Hubinger was also part of the early MATS ecosystem.
Miller was a MATS Fellow.

Nanda and Miller share a research relationship.

Nanda, Hubinger and Perez also appear in the broader AI-safety grantmaking/regranting ecosystem.

So zoom all the way out:

UK IMO 2016/17
↓
COXON ↔ BENTON ↔ NANDA
↓ ↓
ANTHROPIC MATS / ALIGNMENT
↓ ↓
PEREZ ↔ HUBINGER MILLER
↓
PAUSEAI

Then look at what happens in September 2026.

Coxon executes what Miller himself describes as a successful “media strategy.”

Hubinger validates Coxon's central claim.

Perez reveals he'd been discussing Coxon's concerns with him BEFORE the resignation.
Benton backs Coxon's description.
Miller points to the resulting spike in PauseAI UK sign-ups and immediately proposes scaling the mechanism:
“If one previously-unknown Anthropic employee had this much impact, imagine the followup article: ‘4 Anthropic employees and 3 OpenAI employees join Jacob Coxon in quitting.’ Think it through.”

The question is no longer whether these people inhabit the same general AI-safety universe.

They demonstrably do.
Some of these relationships go back TEN YEARS.
The question now is much narrower:
Who knew about Coxon's planned media campaign before it launched, and who helped him execute it?

Sources/receipts below.
https://t.co/LGZLRlK8He
https://t.co/Le479mwehd
https://t.co/5pSW0q0W17

[I Like The Cryptos @LikeCryptos]

@ParkerThayer Jacob does have a lineage in AI but this particular little scheme goes all the way back to his UK IMO team days. I’m putting together a larger scale network for an update but this is what led me down that track.

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@kevinnbass Gonna go ahead and leave this right here

https://t.co/Z8y0a8Ui7z

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@frostbite8768 @ParkerThayer I’m specifically looking into an artificially discovered/amplified story and you’re a bot account that just randomly reappeared specifically just to comment on this topic. It’s not ad hominem lol, it’s common sense and you’re just reaffirming my suspicion.

[I Like The Cryptos @LikeCryptos]

Completely agree! A social manipulation or amplification campaign doesn’t make the underlying safety concerns false. they still have to stand or fall on their merits. That’s actually what interested me about the DSEWiki/HF incidents. The real problems were validation, provenance, and authority boundaries.

[I Like The Cryptos @LikeCryptos]

That’s fair! I don’t think your interpretation is necessarily wrong. I started digging because I don’t like feeling manipulated and suspected there was more context behind what I was seeing. Your responses on this have been pretty thorough and consistent, and I respect you holding your perspective. I’m just following the connections where they lead. There’s a fair bit more behind this connection I’m just organizing it.

[I Like The Cryptos @LikeCryptos]

@jenzhuscott If you want to understand what’s going on with this Coxon affair just connect the dots. Jacob Coxon, Joe Benton, Evan Hubinger, PauseAI/Peter Miller and Neel Nanda. Go from there!

[I Like The Cryptos @LikeCryptos]

If you want to understand what’s going on with this Coxon affair just connect the dots. Jacob Coxon, Joe Benton, Evan Hubinger, PauseAI/Peter Miller and Neel Nanda.

[I Like The Cryptos @LikeCryptos]

@mark_k Pretty much what I’ve found. https://t.co/Z8y0a8Ui7z

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@DavidOndrej1 Not quite, UK AI-Safety orchestrated media push. https://t.co/Z8y0a8Ui7z

[I Like The Cryptos @LikeCryptos]

Something about the Jacob Coxon story doesn't add up.

Coxon appeared essentially overnight, resigned from Anthropic, accused the AI industry of “gambling with our lives,” and his posts exploded to 100M+ views while major media and prominent AI-safety figures amplified him.

But this wasn't simply a researcher spontaneously posting his fears online.

The Wall Street Journal had already interviewed Coxon about his resignation and published an exclusive based on those conversations. Coxon discussed Anthropic's internal Slack, “crunchtime,” “endgame,” and his reasons for leaving.

I somehow found myself checking out Joseph Miller.

Miller is Director of PauseAI UK. Today, on LessWrong, he linked directly to Coxon's resignation and told AI-lab employees that resignations should be deliberately engineered for maximum impact:

“start a social cascade that stigmatises participating in the AI race.”

His instructions?

“You should have a media strategy lined up before you leave, journalists briefed, statements written and edited.”

He then offered:

“If you are in this position, DM me if you want help executing this as effectively as possible.”

And then Miller added this:

“Edit: It seems that Jacob Coxon did have a media strategy and it paid off.”

That's a hell of a statement given what happened.

Miller isn't some anonymous commenter. He runs PauseAI UK, an organization engaged in political advocacy, campaigns and media engagement aimed at stopping or slowing frontier AI development.

And there's another thing.

Miller published an article in July titled “Why PauseAI UK accepts anonymous donations.”

In it, he explicitly acknowledges the obvious concern:

“The main concern here is that PauseAI UK could be an astroturf campaign…”

He discusses hypothetical undisclosed funders including AI investors, AI companies pursuing regulatory capture, investors seeking regulation that affects AI valuations, and others who could financially benefit from portraying AI as extraordinarily dangerous.

In my opinion, this gives us some very specific questions that deserve answers:

Who developed Coxon's media strategy?

Who briefed the journalists?

Who helped write or edit the statements?

Was Coxon in contact with Joseph Miller or PauseAI before his announcement?

And who knew this announcement was coming before the public did?

Because Coxon's story is being presented to millions of people as the spontaneous warning of a terrified AI researcher.

The public deserves to know whether what they actually witnessed was a pre-planned advocacy campaign.

Sources below for as long as they leave them up:

Joseph Miller — Coxon / media-strategy Quick Take:
https://t.co/igo9UGuONu

Joseph Miller — “force multiplier” discussion:
https://t.co/igo9UGuONu

PauseAI UK:
https://t.co/5H8UfjyY04

“Why PauseAI UK accepts anonymous donations”:
https://t.co/WtBeWHIy8w

Wall Street Journal — Coxon exclusive:
https://t.co/Nm0N5vWdXg

[I Like The Cryptos @LikeCryptos]

@iBenGray It’s fun! I just put some more stuff up