The Ex-Congressman Who Says AI Isn't Unstoppable — Brad Carson

Machine Learning Street Talk
Machine Learning Street TalkMay 31, 2026

Why It Matters

Carson’s perspective frames AI governance as a tractable policy challenge, urging lawmakers and firms to impose product‑liability and transparent oversight before unchecked AI deployment reshapes legal and societal norms.

Key Takeaways

  • AI chips remain strategic leverage for U.S. national security.
  • Anthropomorphizing AI risks granting undeserved rights to machines.
  • Regulatory capture threatens effective AI oversight; independent agencies needed.
  • Product liability frameworks should apply to AI misuse and deepfakes.
  • Transparent testing and public accountability are essential for frontier models.

Summary

Former Congressman Brad Carson argues that artificial intelligence is not an unstoppable force; it remains subject to deliberate policy choices and regulatory frameworks. He emphasizes that the United States retains a decisive advantage by controlling the semiconductor chips that power advanced AI, and that this leverage can be used to curb hostile development abroad. Carson warns against the growing tendency to anthropomorphize AI, noting that granting machines perceived personhood erodes legal clarity and opens the door to dangerous rights claims. He critiques the current regulatory environment, describing it as vulnerable to capture by well‑funded tech networks. To counter this, he proposes independent verification bodies—modeled on public‑company accounting oversight—to conduct mandatory testing of frontier models, ensuring transparency without expanding bureaucratic overreach. Carson also stresses that existing product‑liability law should extend to AI systems, holding developers accountable for misuse such as deep‑fake pornography while still assigning responsibility to end users. Key moments include his assertion, “We control the most important part of AI and that is the chips,” and his comparison of AI to a spray‑paint can, arguing it should be regulated as a product, not as speech. He cites the lack of clarity in services like Anthropic’s Claude changes, calling for contractual transparency and consumer protection. Carson also references historical arms‑race cautionary tales, urging a proactive, human‑centered approach to AI governance. The implications are profound: policymakers must craft agencies that can resist capture, enforce rigorous testing, and apply product‑liability standards to emerging AI tools. Industry players will need to adopt transparent practices and anticipate legal responsibilities, while the broader public gains a framework that balances innovation with safety and accountability.

Original Description

Brad Carson was the Army's General Counsel, served two terms in Congress and was Acting Under Secretary of Defense for Personnel and Readiness. He now heads Americans for Responsible Innovation, the AI-policy advocacy group he co-founded. Keith Duggar spends roughly eighty minutes pushing back.
SPONSOR:

Cyber Fund built the Monastery to help founders ship products that were impossible a year ago. Applications for Batch 1 are now open.

Carson's whole case rests on one line: the genie is not out of the bottle. We have pulled dangerous tech back before. Asilomar halted recombinant DNA in 1975, and the West still controls the chips AI runs on. Calling it unstoppable, he says, is the most dangerous idea in the room.
Then Keith drags him somewhere darker. A Palantir heat map scores you 0.73 on whether you are a combatant, and a strike follows. The model is wrong some accepted share of the time, and when it is, nobody answers for it. You cannot court-martial a model, and not even the interpretability researchers can say why it picked you.
Note: after recording, we learned that Americans for Responsible Innovation is backed by EA-aligned philanthropy (not sponsored)

TIMESTAMPS:
00:00:00 From the Pentagon to AI governance
00:04:52 Regulatory capture vs Silicon Valley networks
00:07:56 Transparency and the Claude tier changes
00:09:40 Tort liability when AI tools cause harm
00:13:40 AI is a product, not a person
00:16:01 Children, suicide, and the suicide business
00:19:59 Opaque neural nets and the law of war
00:25:54 Probabilistic targeting and the death of accountability
00:28:47 The arms race fallacy: Asilomar and restraint
00:34:02 Talking to China: track 2 talks and chip leverage
00:39:45 Air power never wins: capital for labour
00:43:29 Anthropic vs the Department of War
00:51:29 Concentration, open source, and brain drain
01:00:18 DeepSeek, Chinese culture, and AI as diplomacy
01:12:25 Upskilling Congress and why public trust matters

REFERENCES:
organization:
[00:02:45] ICRC position on autonomous weapons
[00:05:22] Americans for Responsible Innovation (ARI)
[00:07:20] Andreessen Horowitz (a16z)
[00:43:29] Anthropic
[01:00:18] DeepSeek
[01:03:05] Moonshot AI (Kimi)
[01:16:05] Office of Technology Assessment
other:
[00:03:35] Beneficial AGI 2019 Conference (Future of Life Institute, Puerto Rico)
[00:18:30] Section 230 of the Communications Decency Act
[00:19:59] Lethal Autonomous Weapons (LAWS)
[00:31:35] Strategic Arms Limitation Talks (SALT)
[00:32:28] Asilomar Conference on Recombinant DNA (1975)
[00:39:45] The New Iron Triangle (ARI policy byte)
[00:48:05] Defense Production Act
person:
[00:03:35] Anthony Aguirre
[00:06:48] Dean Ball — Hyperdimensional
[00:23:13] Neel Nanda — mechanistic interpretability
[00:36:02] Jack Clark (Anthropic) on Conversations with Tyler
[00:36:45] Dean Acheson
[00:37:05] Paul Nitze
[00:39:15] Robert Trager — Centre for the Governance of AI
[00:41:55] Giulio Douhet
[01:15:05] Don Beyer (US Congress)
tool:
[00:22:19] Phalanx CIWS
[00:24:50] Palantir Foundry
[01:07:17] Qwen (Alibaba)

ReScript:

Comments

Want to join the conversation?

Loading comments...