Risky Bulletin – “Srsly Risky Biz: Is Claude too woke for war?”

Podcast by Risky Business Media
Episode Date: February 26, 2026
Hosts: Amberly Jack and Tom Uren

Episode Overview

This episode centers on two major cybersecurity policy issues:

1. The disagreement between Anthropic (makers of the Claude AI model) and the US Department of Defense over military AI ethics and control.
2. The ongoing threat posed by the Chinese cyber group Volt Typhoon, despite premature claims of US victory over this adversary.

Through candid analysis and sharp commentary, the hosts explore the nuances of AI restrictions in wartime contexts, who should define “acceptable” AI behavior, and why government messaging matters in cyber defense.

1. Anthropic vs. the Pentagon: Who Sets the Rules for Military AI?

[00:45 – 09:53]

Key Discussion Points

• Background: Anthropic's Claude models have been authorized for classified military use, notably in a recent high-profile raid.
• Tension: Pete Hegseth (Secretary of Defence) has demanded Anthropic remove restrictions on Claude, asserting, "it'll be up to us how we deploy, how we use your technology."
• Anthropic's Stance: CEO Dario Amade states they will not allow Claude “to be used for the mass surveillance of Americans” or “for lethal autonomous weapons.”
• Ethical Paradox: Both hosts note the difficult trade-off between the Pentagon’s operational needs and Anthropic’s ethical guardrails.
• Conceptual Divide:
  • Pentagon View: Considers Claude “a tool and we'll use it how we like.” [03:38]
  • Anthropic View: Sees Claude as edging toward being “an entity, an actual sort of maybe thinking being,” needing clear values to function reliably.
• Claude's Constitution: Anthropic has released a “Claude Constitution” (a set of operational training guidelines) to embed values in the AI’s decision making.

Notable Quotes

• Tom Uren [01:20]:
  "What you do in a Defense Department is sometimes you blow things up and kill people. That's not the way you want normal AI models to be trained. So you've got to get those safeguards out of those systems."
• Tom Uren [03:38]:
  "I think that what the Department of Defense or war really wants is not Claude with Claude's standard constitution, but Claude with like a warrior ethos."
• Amberly Jack [08:22]:
  "Congress needs to take a kind of a mentor role here and, you know, almost to stop the kids from fighting too much and getting out of hand."

Central Questions Raised

• Should AI training for the military include or exclude value-based constraints?
• Who should decide what values get embedded: defense officials, AI companies, or Congress?
• Are “set and forget” legal frameworks even possible in such a rapidly changing technological environment?
  • Tom Uren [08:03]:
    "I think there's very few set and forget laws that can be implemented."

Congressional Role

• Both hosts agree Congress should define transparency, oversight, and the guiding principles for military AI use.
• Uren warns against leaving these decisions solely to industry or the military:
  • Tom Uren [06:00]:
    "Congress should be getting involved...to make sure that not only are AI companies making models that are appropriate for military use, but that that military use is, is consistent with American values."

2. Volt Typhoon: The Persistent Threat and the Perils of Declaring Victory

[09:53 – 15:45]

Key Discussion Points

• Volt Typhoon’s Activity: A Chinese state-backed group infiltrating US critical infrastructure, likely as pre-positioning for potential conflict (notably concerning Taiwan).
• US Government Messaging: In July 2025, officials claimed “mission accomplished” against Volt Typhoon, attributing success to private sector support.
  • Tom’s Skepticism:
    "This would be the only time a government has ever declared, yes, total victory against a cyber threat...because the entire history of cyber threats is that they don't go away, they just evolve." [10:35]
• Update: Dragos, a leading cybersecurity firm, confirms Volt Typhoon remains active and embedded in US systems ([12:10]).
• Communication Impact: Overstated government victory can breed complacency in private sector defenders and undermine long-term preparedness.

Notable Quotes & Moments

• Tom Uren [12:14]:
  "The problem with declaring victory, like it looks good in the short term, but...if you say we've totally won, everyone's going to go, well, why should I bother doing anything?"

• Contrast with Salt Typhoon:

  • US officials maintain open communication and urge cooperation about another Chinese group (“Salt Typhoon”), highlighting the inconsistency.
  • Amberly Jack [14:17]:
    "The strategy with Salt Typhoon seems like a pretty good way to get these organizations and these telcos to cooperate. And that's kind of exactly what should be happening with Volt Typhoon."

• Necessity of Balanced Messaging:

  • Tom Uren [14:38]:
    "You don't want to say Vault Typhoon is the worst thing ever... I don't think you want to say nothing to see here... It's a contested area where you need constant effort to push back against what the Chinese are doing."

• Memorable Close:

  • Amberly Jack [15:31]:
    "Pretty sure we all as kids learn very quickly the lesson that pretending problems don't exist doesn't mean that they go away."
  • Tom Uren [15:45]:
    "I'm not sure that we did all learn that as kids. I find that there's many adults who employ that strategy still."

3. Takeaways & Emerging Themes

AI for War: A Struggle Over Values, Power, and Oversight

• Military AI is at a crossroads: the need for effectiveness versus the ethical guardrails that define American values.
• There’s a notable tension between treating advanced AI as a “tool” versus an “entity” requiring explicit value alignment.
• Democratic oversight (via Congress) is necessary, but practical and philosophical challenges abound.

Cyber Threats: Evolution, Not Erasure

• Government pronouncements of total victory are not only inaccurate but counterproductive.
• Open, honest threat communication—like that used for Salt Typhoon—better engages private sector defenders.

4. Important Segment Timestamps

• [00:45 – 09:53]: Anthropic vs. Pentagon on AI restrictions
  • Pentagon’s demands vs. Anthropic’s AI guardrails
  • Claude’s “constitution” as a model for embedded AI values
  • Congressional responsibility in AI governance
• [09:53 – 15:45]: Volt Typhoon’s resilience and policy pitfalls
  • The myth of “mission accomplished” in cybersecurity
  • How government messaging affects national cyberspace resilience
  • The case for balanced, ongoing engagement over cybersecurity threats

5. Closing

The hosts deliver a timely, cogent discussion about who gets to set the rules—both in the domain of military AI and national cyber defense. The message: Real world security requires balancing philosophical, ethical, and practical considerations and clear, honest communication.

(End of summary. You can read Tom's full newsletter on the Risky Biz website for more details.)