Amos Toh
@amostoh.bsky.social
senior counsel, Brennan Center for Justice. interested in all the ways money in tech helps and hurts. he/him, 🇸🇬 in 🇺🇸
Pinned
Amos Toh
@amostoh.bsky.social
· Mar 11
The Business of Military AI
The Pentagon has been spending tens of billions of dollars to adopt new technologies at breakneck speed. Without oversight and safeguards, military applications of artificial intelligence could jeopar...
www.brennancenter.org
1/ As the US continues AI-enabled strikes on Iran, new @brennancenter.org research from @emileayoub.bsky.social and me examines how the military’s investments in the technology have been building to this moment.
Congress must step up and reckon with AI’s dangers to life, liberty, and democracy 🧵:
Congress must step up and reckon with AI’s dangers to life, liberty, and democracy 🧵:
AI companies and the partners they handpicked triggered the recent wave of cyberattacks while tinkering with experimental models, pushing their limits to see how dangerous they could get.
My new @brennancenter.org essay explains how lawmakers should regulate risky uses of AI the public doesn't see👇
My new @brennancenter.org essay explains how lawmakers should regulate risky uses of AI the public doesn't see👇
The Conversation We Need to Have About Regulating AI
Safeguards must address not only public releases but also risky uses of advanced models behind closed doors.
www.brennancenter.org
October 5, 2026 at 4:34 PM
AI companies and the partners they handpicked triggered the recent wave of cyberattacks while tinkering with experimental models, pushing their limits to see how dangerous they could get.
My new @brennancenter.org essay explains how lawmakers should regulate risky uses of AI the public doesn't see👇
My new @brennancenter.org essay explains how lawmakers should regulate risky uses of AI the public doesn't see👇
Senator Hawley just disclosed that Sam Altman turned down his invitation to testify at the hearing. (None of the AI developers are testifying.)
Live video here:
www.congress.gov/event/119th-...
Live video here:
www.congress.gov/event/119th-...
September 30, 2026 at 7:16 PM
Senator Hawley just disclosed that Sam Altman turned down his invitation to testify at the hearing. (None of the AI developers are testifying.)
Live video here:
www.congress.gov/event/119th-...
Live video here:
www.congress.gov/event/119th-...
🧵 Congress is holding its first hearing today on the rogue AI cyberattacks.
METR, which surfaced important facts about the Hugging Face hack, is testifying.
But OpenAI limited its investigation, meaning that there could be a lot more we don't know.
Here are 7 questions lawmakers should ask👇:
METR, which surfaced important facts about the Hugging Face hack, is testifying.
But OpenAI limited its investigation, meaning that there could be a lot more we don't know.
Here are 7 questions lawmakers should ask👇:
How Congress Should Investigate Threat from Rogue AI Agents
The government must evaluate systemic failures with an eye toward establishing meaningful regulation.
www.brennancenter.org
September 30, 2026 at 3:48 PM
🧵 Congress is holding its first hearing today on the rogue AI cyberattacks.
METR, which surfaced important facts about the Hugging Face hack, is testifying.
But OpenAI limited its investigation, meaning that there could be a lot more we don't know.
Here are 7 questions lawmakers should ask👇:
METR, which surfaced important facts about the Hugging Face hack, is testifying.
But OpenAI limited its investigation, meaning that there could be a lot more we don't know.
Here are 7 questions lawmakers should ask👇:
@nytimes.com found that OpenAI executives pressured employees to fast-track model testing despite safety concerns.
All the more reason for Congress to investigate how decisions to prioritize speed over safety contributed to the hacking incidents. Here's how 👇
www.brennancenter.org/our-work/res...
All the more reason for Congress to investigate how decisions to prioritize speed over safety contributed to the hacking incidents. Here's how 👇
www.brennancenter.org/our-work/res...
OpenAI Ignored Employees’ Warnings About Safely Testing A.I. Models
Employees and security researchers said that they had cautioned the company on safely testing its A.I. models and strengthening its corporate infrastructure, but that OpenAI did not listen.
www.nytimes.com
September 29, 2026 at 8:38 PM
@nytimes.com found that OpenAI executives pressured employees to fast-track model testing despite safety concerns.
All the more reason for Congress to investigate how decisions to prioritize speed over safety contributed to the hacking incidents. Here's how 👇
www.brennancenter.org/our-work/res...
All the more reason for Congress to investigate how decisions to prioritize speed over safety contributed to the hacking incidents. Here's how 👇
www.brennancenter.org/our-work/res...
Rogue AI or rogue AI industry?
My latest for @brennancenter.org explains how Congress should investigate AI companies for the recent wave of hacking incidents.
Lawmakers should focus on whether the race to build ever more powerful AI sidelined digital security:
My latest for @brennancenter.org explains how Congress should investigate AI companies for the recent wave of hacking incidents.
Lawmakers should focus on whether the race to build ever more powerful AI sidelined digital security:
How Congress Should Investigate Threat from Rogue AI Agents
The government must evaluate systemic failures with an eye toward establishing meaningful regulation.
www.brennancenter.org
September 29, 2026 at 4:48 PM
Rogue AI or rogue AI industry?
My latest for @brennancenter.org explains how Congress should investigate AI companies for the recent wave of hacking incidents.
Lawmakers should focus on whether the race to build ever more powerful AI sidelined digital security:
My latest for @brennancenter.org explains how Congress should investigate AI companies for the recent wave of hacking incidents.
Lawmakers should focus on whether the race to build ever more powerful AI sidelined digital security:
My latest for @brennancenter.org on why the Trump admin’s moves on AI and cybersecurity confound rather than clarify.
Ban on foreign access to Claude’s most advanced models underscores the challenges of governing when the admin’s instincts are anti-regulatory:
www.brennancenter.org/our-work/ana...
Ban on foreign access to Claude’s most advanced models underscores the challenges of governing when the admin’s instincts are anti-regulatory:
www.brennancenter.org/our-work/ana...
Trump’s Chaotic AI and Cybersecurity Policy
The administration’s gutting of agencies, contradictory impulses on regulating AI, and susceptibility to outside interests undermine trust in its capacity to oversee artificial intelligence risks.
www.brennancenter.org
June 17, 2026 at 5:44 PM
My latest for @brennancenter.org on why the Trump admin’s moves on AI and cybersecurity confound rather than clarify.
Ban on foreign access to Claude’s most advanced models underscores the challenges of governing when the admin’s instincts are anti-regulatory:
www.brennancenter.org/our-work/ana...
Ban on foreign access to Claude’s most advanced models underscores the challenges of governing when the admin’s instincts are anti-regulatory:
www.brennancenter.org/our-work/ana...
Reposted by Amos Toh
It is now crystal clear who is obstructing reauthorization. Cornyn & other reform opponents would rather see Sec. 702 expire than allow reforms to protect Americans’ privacy. Sen. Wyden offered an easy path forward for extending the law. Reform opponents would prefer sunset. 2/2
June 10, 2026 at 9:20 PM
It is now crystal clear who is obstructing reauthorization. Cornyn & other reform opponents would rather see Sec. 702 expire than allow reforms to protect Americans’ privacy. Sen. Wyden offered an easy path forward for extending the law. Reform opponents would prefer sunset. 2/2
Reposted by Amos Toh
Senator Cornyn just blocked a 9-month extension of Section 702 with a warrant requirement for backdoor searches. He then blocked a 5-week extension with no warrant requirement, just basic transparency measures. 1/2
June 10, 2026 at 9:20 PM
Senator Cornyn just blocked a 9-month extension of Section 702 with a warrant requirement for backdoor searches. He then blocked a 5-week extension with no warrant requirement, just basic transparency measures. 1/2
Reposted by Amos Toh
In the guise of an AI powered truth tribunal, Peter Thiel and Balaji Srinivasan are backing a new platform to supercharge attacks on the press and promote "justice" without judges or juries. Don't buy it: www.codastory.com/polarization...
Peter Thiel is building a parallel justice system — Powered by AI - Coda Story
His investment in Objection.ai points to a new model: private investigations, AI verdicts, and accountability mechanisms that operate outside democratic institutions.
www.codastory.com
April 20, 2026 at 10:46 PM
In the guise of an AI powered truth tribunal, Peter Thiel and Balaji Srinivasan are backing a new platform to supercharge attacks on the press and promote "justice" without judges or juries. Don't buy it: www.codastory.com/polarization...
Reposted by Amos Toh
The risks of military AI "are already existential," says @amostoh.bsky.social -- "both for civilians swept up in the violence and destruction of AI-enabled wars and troops that have to live with the consequences of potentially unsafe weapons they can’t control.” theintercept.com/2026/05/01/e...
Musk Warns of Killer AI — While He and the Rest of Silicon Valley Cash In on AI That Kills
In his lawsuit against OpenAI, Elon Musk evoked a “Terminator” scenario. He said nothing about the people AI is already killing.
theintercept.com
May 2, 2026 at 5:35 PM
The risks of military AI "are already existential," says @amostoh.bsky.social -- "both for civilians swept up in the violence and destruction of AI-enabled wars and troops that have to live with the consequences of potentially unsafe weapons they can’t control.” theintercept.com/2026/05/01/e...
Reposted by Amos Toh
What happens when the US military races to implement AI in war?
I spoke with @amostoh.bsky.social, senior counsel at the @brennancenter.org, about what really goes on in these massive, secretive US military-tech contracts.
Listen to the conversation here: www.themaybe.org/podcast/comp...
I spoke with @amostoh.bsky.social, senior counsel at the @brennancenter.org, about what really goes on in these massive, secretive US military-tech contracts.
Listen to the conversation here: www.themaybe.org/podcast/comp...
Computer Says Kill: A License for Unlimited War w/ Amos Toh
Who’s in control of calling the shots on AI in war— the military or the tech companies that they’re licensing their tech from?
www.themaybe.org
May 1, 2026 at 9:22 PM
What happens when the US military races to implement AI in war?
I spoke with @amostoh.bsky.social, senior counsel at the @brennancenter.org, about what really goes on in these massive, secretive US military-tech contracts.
Listen to the conversation here: www.themaybe.org/podcast/comp...
I spoke with @amostoh.bsky.social, senior counsel at the @brennancenter.org, about what really goes on in these massive, secretive US military-tech contracts.
Listen to the conversation here: www.themaybe.org/podcast/comp...
@theinformation.com just revealed key details of Google’s deal with the Pentagon to deploy Gemini in warfare.
Charlie Bullock and I explain how loopholes in the deal pave the way for Gemini’s use in mass domestic surveillance and fully autonomous weapons - Google’s supposed red lines.
Charlie Bullock and I explain how loopholes in the deal pave the way for Gemini’s use in mass domestic surveillance and fully autonomous weapons - Google’s supposed red lines.
April 28, 2026 at 6:01 PM
@theinformation.com just revealed key details of Google’s deal with the Pentagon to deploy Gemini in warfare.
Charlie Bullock and I explain how loopholes in the deal pave the way for Gemini’s use in mass domestic surveillance and fully autonomous weapons - Google’s supposed red lines.
Charlie Bullock and I explain how loopholes in the deal pave the way for Gemini’s use in mass domestic surveillance and fully autonomous weapons - Google’s supposed red lines.
critical intervention from @stevenfeldstein.bsky.social, on how Big Tech is resetting the terms of warfare. Steve foreshadowed the use of "good enough" drones as geopolitical chips - and look at where we are now
foreignpolicy.com/2025/11/17/d...
foreignpolicy.com/2025/11/17/d...
April 23, 2026 at 2:59 PM
critical intervention from @stevenfeldstein.bsky.social, on how Big Tech is resetting the terms of warfare. Steve foreshadowed the use of "good enough" drones as geopolitical chips - and look at where we are now
foreignpolicy.com/2025/11/17/d...
foreignpolicy.com/2025/11/17/d...
Reposted by Amos Toh
"The Capture of the Congressional Intelligence Committees" is forthcoming from the UC Irvine Law Review! It argues that Congress's intelligence oversight committees have become the intelligence community's biggest boosters on Capitol Hill.
papers.ssrn.com/sol3/papers....
papers.ssrn.com/sol3/papers....
February 16, 2026 at 3:42 PM
"The Capture of the Congressional Intelligence Committees" is forthcoming from the UC Irvine Law Review! It argues that Congress's intelligence oversight committees have become the intelligence community's biggest boosters on Capitol Hill.
papers.ssrn.com/sol3/papers....
papers.ssrn.com/sol3/papers....
Reposted by Amos Toh
Very excited to share the cover of my forthcoming book, OUTED: LBJ’S CONFIDANT AND THE ARREST THAT TRANSFORMED A PRESIDENCY.
Coming October 13 from University of Chicago Press!
press.uchicago.edu/ucp/books/bo...
Coming October 13 from University of Chicago Press!
press.uchicago.edu/ucp/books/bo...
April 1, 2026 at 4:30 PM
Very excited to share the cover of my forthcoming book, OUTED: LBJ’S CONFIDANT AND THE ARREST THAT TRANSFORMED A PRESIDENCY.
Coming October 13 from University of Chicago Press!
press.uchicago.edu/ucp/books/bo...
Coming October 13 from University of Chicago Press!
press.uchicago.edu/ucp/books/bo...
SCOTUS may well affirm the constitutional guarantee of birthright citizenship.
But how we got to the point where this is even up for debate speaks volumes about the Supreme Court’s failures.
Must-read 🧵 from @tomtmwolf.bsky.social 👇
But how we got to the point where this is even up for debate speaks volumes about the Supreme Court’s failures.
Must-read 🧵 from @tomtmwolf.bsky.social 👇
1. SCOTUS is going to have a hard time ruling for Trump on birthright. He did himself no favors today.
2. Our birthright citizenship rule is a universal one, designed to prevent Trump from doing what he's trying to do now -- deciding who's American and who isn't.
3. SCOTUS? Still a major problem.🧵
2. Our birthright citizenship rule is a universal one, designed to prevent Trump from doing what he's trying to do now -- deciding who's American and who isn't.
3. SCOTUS? Still a major problem.🧵
April 1, 2026 at 7:07 PM
SCOTUS may well affirm the constitutional guarantee of birthright citizenship.
But how we got to the point where this is even up for debate speaks volumes about the Supreme Court’s failures.
Must-read 🧵 from @tomtmwolf.bsky.social 👇
But how we got to the point where this is even up for debate speaks volumes about the Supreme Court’s failures.
Must-read 🧵 from @tomtmwolf.bsky.social 👇
Reposted by Amos Toh
Glad to see you quoted @amostoh.bsky.social, but there shd be more emphasis on the fact that the tech elites placing the bets worked hard to make them pay off. It's not just a bet, it was an aim, enaced with millions of #$S in lobbying & shaping policy over a decade. www.cambridge.org/core/journal...
March 28, 2026 at 9:56 AM
Glad to see you quoted @amostoh.bsky.social, but there shd be more emphasis on the fact that the tech elites placing the bets worked hard to make them pay off. It's not just a bet, it was an aim, enaced with millions of #$S in lobbying & shaping policy over a decade. www.cambridge.org/core/journal...
A clear-eyed rebuke of executive overreach and a decisive victory for the First Amendment.
But the administration shows no signs of stopping its crusade against "woke AI," the data brokers' loophole is still wide open, and thousands are dying or displaced in an AI-fueled war.
Where is Congress?
But the administration shows no signs of stopping its crusade against "woke AI," the data brokers' loophole is still wide open, and thousands are dying or displaced in an AI-fueled war.
Where is Congress?
"A federal judge in San Francisco blocked a Pentagon order Thursday labeling artificial intelligence company Anthropic a national security risk, saying officials had likely violated the law and retaliated against the firm for speaking publicly about how it wanted its technology to be used."
Judge blocks Pentagon order branding Anthropic a national security risk
Anthropic argued that the Trump administration was punishing it for speaking about the risks of its AI technology amid a bitter dispute with the Pentagon.
www.washingtonpost.com
March 27, 2026 at 1:35 AM
A clear-eyed rebuke of executive overreach and a decisive victory for the First Amendment.
But the administration shows no signs of stopping its crusade against "woke AI," the data brokers' loophole is still wide open, and thousands are dying or displaced in an AI-fueled war.
Where is Congress?
But the administration shows no signs of stopping its crusade against "woke AI," the data brokers' loophole is still wide open, and thousands are dying or displaced in an AI-fueled war.
Where is Congress?
We at @brennancenter.org found that Palantir’s defense revenue surged over the last decade. 2025 was their most profitable year ever.
Numbers can mislead. But sometimes they just don’t lie.
www.brennancenter.org/our-work/res...
Numbers can mislead. But sometimes they just don’t lie.
www.brennancenter.org/our-work/res...
March 23, 2026 at 3:14 AM
We at @brennancenter.org found that Palantir’s defense revenue surged over the last decade. 2025 was their most profitable year ever.
Numbers can mislead. But sometimes they just don’t lie.
www.brennancenter.org/our-work/res...
Numbers can mislead. But sometimes they just don’t lie.
www.brennancenter.org/our-work/res...
Reposted by Amos Toh
Great post, particularly around the concept of “decision.” Decision has been used to describe what computers do going back at least to Shannon? And now it’s slipping to suggest that if an LLM makes a “decision” then a human has not. But a human is still making a decision about how to decide.
March 22, 2026 at 4:22 AM
Great post, particularly around the concept of “decision.” Decision has been used to describe what computers do going back at least to Shannon? And now it’s slipping to suggest that if an LLM makes a “decision” then a human has not. But a human is still making a decision about how to decide.
Reposted by Amos Toh
”debased” (adj.): 👇
Overnight: The Trump administration is making a renewed push to deport Kilmar Abrego Garcia, telling a judge they've sorted out logistics to send him to Liberia (and rejected his proposal to go to Costa Rica)
assets.bwbx.io/documents/us...
assets.bwbx.io/documents/us...
March 21, 2026 at 11:19 PM
”debased” (adj.): 👇
Illuminating story from @sheeraf.bsky.social @nytimes.com, situating the Iran war in Silicon Valley’s decade-long push to transform warfare.
Who stands to lose the most when industry bets big on war? Civilians and taxpayers.
@brennancenter.org explains why: www.brennancenter.org/our-work/res...
Who stands to lose the most when industry bets big on war? Civilians and taxpayers.
@brennancenter.org explains why: www.brennancenter.org/our-work/res...
Silicon Valley Bet on War. The Bets Are Paying Off.
www.nytimes.com
March 19, 2026 at 1:31 PM
Illuminating story from @sheeraf.bsky.social @nytimes.com, situating the Iran war in Silicon Valley’s decade-long push to transform warfare.
Who stands to lose the most when industry bets big on war? Civilians and taxpayers.
@brennancenter.org explains why: www.brennancenter.org/our-work/res...
Who stands to lose the most when industry bets big on war? Civilians and taxpayers.
@brennancenter.org explains why: www.brennancenter.org/our-work/res...
Reposted by Amos Toh
The Dept. of Defense's headlong rush to adopt AI means overreliance on just a handful of tech companies and weak oversight, says the Brennan Center's @amostoh.bsky.social.
Silicon Valley Bet on War. The Bets Are Paying Off.
www.nytimes.com
March 18, 2026 at 2:45 PM
The Dept. of Defense's headlong rush to adopt AI means overreliance on just a handful of tech companies and weak oversight, says the Brennan Center's @amostoh.bsky.social.
Reposted by Amos Toh
This is an incredibly valuable report and I recommend everyone read it and get a primer on how far down the rabbit hole of AI warfare we've traveled so far.
1/ As the US continues AI-enabled strikes on Iran, new @brennancenter.org research from @emileayoub.bsky.social and me examines how the military’s investments in the technology have been building to this moment.
Congress must step up and reckon with AI’s dangers to life, liberty, and democracy 🧵:
Congress must step up and reckon with AI’s dangers to life, liberty, and democracy 🧵:
The Business of Military AI
The Pentagon has been spending tens of billions of dollars to adopt new technologies at breakneck speed. Without oversight and safeguards, military applications of artificial intelligence could jeopar...
www.brennancenter.org
March 16, 2026 at 4:59 PM
This is an incredibly valuable report and I recommend everyone read it and get a primer on how far down the rabbit hole of AI warfare we've traveled so far.
Reposted by Amos Toh
Note how the AI "recommendations" are completely obscured with little to no ability to actually verify or trace their outputs. This is what we mean when we say the distinctions between DSS and AWS are superficial in practice, especially when operators are given seconds to approve.
🇺🇸⚡️🇮🇷— The Pentagon uses Palantir’s AI system, Maven, when planning strikes on Iran.
Commanders select a target, the system suggests the best munitions, and a live feed shows the strike in real time.
Commanders select a target, the system suggests the best munitions, and a live feed shows the strike in real time.
March 14, 2026 at 10:26 PM
Note how the AI "recommendations" are completely obscured with little to no ability to actually verify or trace their outputs. This is what we mean when we say the distinctions between DSS and AWS are superficial in practice, especially when operators are given seconds to approve.