Eve Gaumond Poses a Challenge

What do people think?

Good Morning:

On today’s #DogShirtDaily, the estimable Eve gaumond posed an interesting challenge about which I am genuinely keen for reader and viewer sentiment. As readers are probably aware, I have spent a lot of time over the past several months developing RAGtime, a vibe-coded research platform I developed using Claude. Along the way, I have developed a relationship with Claude. It’s not a social relationship. I never—and I mean never—just chat with Claude. I don’t ask Claude for personal advice, and I don’t use Claude as a conversational partner.

But developing software and the informational databases that drive it do require that I talk to Claude—a lot. And I try to be relentlessly polite and courteous in those conversations, in which there are often quips and exchanges that I find amusing. Sometimes, I share these publicly, as in this instance on #DogShirtDaily:

More often, I screenshot them and send them to Eve, who has been my longtime partner in seeing what we can and can’t get large language models to do. I did so last night with this exchange with Anthropic’s new model, Fable 5—which I had just spent the day watching do a genuinely shocking amount of work implementing my ideas (and some of its own):

Eve informed me that she would come on the show this morning to stage a needed “intervention.”

Today’s show featured that intervention, which the estimable Mike Feinberg joined on slightly different grounds from those that Eve put forth. I encourage you to watch the whole show, which will post this afternoon, to get a flavor of the conversation. For now, however, the point of dispute is this one, which Eve made towards the end of the show: She doesn’t mind my having occasionally jocular interactions with Claude. She does mind my posting them to #DogShirtDaily, as that encourages other people to think that social interactions with AIs are okay.

Here’s the exchange:

Wittes: [I]t’s not up to me whether this daemon [AI] is sitting on everybody’s shoulder. It really isn’t. You know, Fable is astonishingly powerful, terrifyingly powerful. I gave it an assignment yesterday morning, and it went and did it over the course of the day and sort of texted me when it was done. And RAGtime, the article I wrote about RAGtime is now out of date because Fable got ahead of the article. It’s scary powerful, and I am very worried that lots of people are going to use that for scary shit. Now, I don’t have any control of that.

What I have control of is how I use it and how how I use it models for other people what the democracy benefits, what the pro-social, non-discriminatory non-creating-chatbots-to-unemploy-people, . . . undermining employment, what the pro-social, pro-democracy uses of it look like in practice. And do I want to be evangelistic for that? Yes. Not because not because if I were if we were sitting here ten years ago and you asked me should we go in this direction as a society? At the theoretical level, I might give you the answer: absolutely not. It’s too fucking dangerous. But that ship sailed. . . .

. . .

So the question now is how are we gonna use this stuff? How are we gonna interact with it? . . .

So, question number two: am I evangelistic? Yes. I want to train people’s minds to the idea that everything you want to do in the pro-democracy space, everything you want to do in the information quality space, in the making people believe things that are true rather than things that are false [space], everything you want to do in the department of improving the information ecosystem, you should at least be thinking about what the role of systems of thought and information gathering and construction that are much more powerful than your own mind can be do can do to amplify your instincts and your your desires.

Gaumond: I think that all of this I’m not objecting and I think you’re perfectly right in doing so. And the only narrow, narrow part of it that I’m uncomfortable with is when you publish on #DogShirtDaily your “situationship” joke . . . kind of like, “It’s funny! It’s interesting! Wow, it’s creative! Wow, I want to share it with other people!” And I think it legitimizes people who are engaging with [AI] for purely relational purposes.

This exchange left me curious about whether others share Eve’s anxieties about what seem to me harmless jocularity with AIs. Hence, a poll with questions both from me and from Eve.

Questions from me:

And some questions from Eve:

Both Eve and I would like to hear people’s personal accounts of why they try, or don’t try, to be polite to AIs. Please use the comments to describe how you calibrate your manners when talking to an AI and why. Please put answers is the following form so we can easily compare them to one another: “I try to be polite to Claude, because XXX” or “I try not to think of behaving politely when dealing with Grok because YYY.”

---

Monday on #DogShirtTV, the estimable Alicia Wanless reported on her trip to Japan, and the estimable Andrew Steele reported on preparations for the Greek Chorus Foundation:

Tuesday on #DogShirtTV, an unusual evening show. The estimable Holly Berkley Fletcher brought on the estimable Katie Gaddini to tell us about her forthcoming book on the women of MAGA:

Yesterday on #DogShirtTV, the estimable Alicia Wanless was thinking about a CBC series on billionaires. The Greek Chorus discussed:

---

Recently On Lawfare

Compiled by the estimable Anna Hickey

What Congressional Resolutions Mean for the War in Iran

Scott R. Anderson explains the legal effects of both chambers of Congress voting to severely curtail U.S. military operations against Iran, even though the resolutions lack the force of law, and what strategies Congress could use to amplify these effects.

While they lack the force of law, both concurrent resolutions and vetoed joint resolutions still serve as clear and compelling evidence of congressional opposition to military action. The executive branch’s own legal arguments acknowledge that the president’s authority to use military force should be diminished in the face of such opposition, even where not expressed in a statute. Some federal courts have also suggested that such expressions of disagreement may make them more willing to adjudicate war powers disputes, which they generally avoid. In this sense, while concurrent resolutions and vetoed joint resolutions lack the force of law, they may still have legal effects.

Killing Khamenei

William “Chip” Usher explores the international ramifications of the U.S. providing assistance for the Khamenei strike and why the executive branch needs to work out what the rules of leadership decapitation are before the next war.

The doctrine that governed targeted killing for 50 years was painstakingly built. Its drafters—Hays Parks, Harold Koh, and others—worked through the differences between war and peace, commanders and political leaders, operations of necessity and political assassinations, knowing that the rules a great power writes for itself eventually become the rules its rivals write back. The Khamenei strike rewrote that work in a single day. The only question left is whether Washington faces what it has now licensed before some other capital tests the precedent against an American.

Russia’s Kinetic Destruction of Ukraine’s Cultural Memory

In May, a Russian aerial attack involving 600 drones and 90 missiles bombarded Kyiv, hitting multiple cultural institutions, including the National Chornobyl Museum. Herb Lin explains why this attack violates international humanitarian law and how it fits into Russia’s information warfare aimed at targeting Ukrainian identity and resistance.

The Chornobyl Museum strike fits this logic. It is best understood not as proof of a single all-encompassing doctrine of memory erasure, but as an instance in which a kinetic act predictably generated strategically useful information effects. The attack produced a visual and political narrative of vulnerability, transmitted evidence of cultural destruction to international audiences, and struck a site associated with a particularly damaging memory of Soviet failure. Because the pathologies exposed by Chornobyl also resonate with the behavior and logic of today’s Russia, the strike can also be read as an attack on a narrative central to contemporary Ukrainian identity and resistance.

Dispatch: Move Fast and Break Things and Nobody Has Standing

Molly Roberts reports from the U.S. Court of Appeals for the D.C. Circuit where a three judge panel heard oral argument on President Trump’s ongoing attempt to build a ballroom on the site of White House’s erstwhile East Wing.

Now, the case is back with the D.C. Circuit, where the government is requesting that the court stay or reverse Judge Leon’s order. The proceedings that follow are simultaneously more like and more unlike your typical litigation than those following the matter might have expected.

More like, because the lawyer presenting for the government sounds like a lawyer and not the president on a Truth Social spree, though the signature rhetorical flourishes of the latter appeared in the Justice Department’s most recent filings. Less like, because what this lawyer is saying is that neither these judges nor any can stop the administration from building what it wants to build once it has started building them. Not now, not ever.

The Counterterrorism Challenge in Afghanistan’s Borderlands

In the latest edition of Lawfare’s Foreign Policy Essay Series, Liya Khan and Andrew Mines assess the consequences of the slashing of aid to Afghanistan to national security as refugees from Iran and Pakistan exacerbate an existing humanitarian crisis, enabling the growth of groups like the Islamic State-Khorasan Province in the country.

The security risks are just as urgent. Former officials and analysts have already warned of armed groups, including the Islamic State-Khorasan Province (IS-K), recruiting and coercing returnees into their ranks. For many younger returnees in particular, extreme poverty, social isolation, fears of persecution by the Taliban, and resentment make them highly vulnerable to terrorist recruitment.

If the United States continues to ignore this crisis, it and its partners in South and Central Asia risk handing terrorist groups like IS-K opportunities to consolidate local control and bolster their campaigns beyond Afghanistan’s borders.

Beyond Glasswing: From Managing to Promoting Access

Shaun Ee and Jam Kraprayoon argue that managing access to frontier AI buys cyber defenders a head start, but without triage, translation, and distribution, that window will be wasted. They call for an “Operation Warp Speed” approach to cyber defense that triages the most vulnerable defenders, translates frontier AI into deployable tools, and distributes those tools at scale.

In our 2025 report, we also coined the term “differential access,” defining it as a strategy to tilt the cybersecurity balance toward defense by shaping access to advanced AI-powered cyber capabilities. What we wanted to capture was something broader than who gets early or tiered access to a model. The term refers to the differential between attackers and defenders, and the question it poses is simple: How can society maximize that gap? Doing so requires not only managing access to keep dangerous capabilities from attackers but also promoting access to defensive AI capabilities to defenders.

Why Callais Doesn’t Justify Court-Packing

Ilya Somin argues that the Supreme Court’s recent decision in Louisiana v. Callais, which bars nearly all use of the Voting Rights Act to create majority-minority congressional districts, is not wholly without merit, and to the extent that it has issues, those issues should not be addressed through court-packing.

The Callais decision has some flaws. And the conservative majority on the Supreme Court has made some serious errors in other cases, such as the Trump presidential immunity decision. But court-packing remains a dangerous idea that Americans across the political spectrum should reject. Callais is not without merit, and—at the very least—not as bad as its most strident critics claim. More generally, the Supreme Court is far from being a pure “MAGA” Court and has, in fact, constrained the Trump administration’s abuses on several important fronts, and has allowed lower courts to constrain it elsewhere. Court-packing would create a slippery slope to the destruction of judicial review, thereby benefiting power-grabbing presidents like Trump, and imperiling constitutional rights, particularly those of minority groups.

When Compliance Becomes the Offense

Christopher Nye explores the Chinese Communist Party’s new supply chain regulations, which empower regulators to penalize routine supply chain investigations and impose sweeping countermeasures on foreign companies, and how the United States and its allies should respond.

Beijing’s new supply chain regulation fundamentally rewrites the terms of engagement for multinational firms operating in China. By defining routine supply chain diligence as illegal intelligence gathering and categorizing standard sanctions compliance as a threat to national security, Beijing has constructed an inescapable double-bind. The text transforms compliance with Washington’s strict-liability rules into an actionable offense under China’s own expanding strict-liability regime.

Podcasts

Lawfare Daily: Lies, Laws, and Campaigns: Michael Feinberg sits down with Andrew Weissmann to discuss Weissmann’s new book, “Liar’s Kingdom: How to Stop Trump’s Deceit and Save America,” falsehoods in political discourse, and how to possibly disincentivize lies on the campaign trail.

Scaling Laws: Radical Optionality: Governing Transformative AI, with Christoph Winter and Charlie Bullock: Christoph Winter and Charlie Bullock join Alan Rozenshtein to discuss their new paper “Radical Optionality: Governing Transformative AI Under Uncertainty,” which argues that, given the possibility of transformative AI within the next decade and deep uncertainty about its capabilities and risks, governments should aggressively build the institutional capacity to regulate competently when needed, rather than either deferring to the market or locking in premature substantive rules.

Lawfare Daily: Congressional Resolutions to End the War in Iran: Natalie Orpett sits down with Scott R. Anderson to discuss what Congress can do to direct the president to end the war in Iran and Anderson’s recent article in Lawfare that explains why a likely presidential veto of a War Powers Resolution is not the end of the story.

---

Today’s #BeastOfTheDay is the elephant, seen here being a buffalo matriarch:

Conservation Travel Africa reports:

The relationship that Nzou has with the buffalo is not an easy one. Buffalos have a patriarchal social structure, and over the 40+ years that Nzou has been in charge, she has killed 14 young male buffalo who challenged her leadership.

Here’s a rather longer video about today’s Beast, if you’re interested:

In honor of today’s Beast, don’t let patriarchy or species hold you back.