Ben Harack

Used cars are cheap because of new cars. Day-old bread it half price because of fresh bread.

This has been a housing policy post.

Screenshot from the French National Strategic Review 2025, page 34: "34 1.2 Vital interests 171. It is the sole responsibility of the President of the Republic to assess what these interests are. The choice to maintain deliberate ambiguity, reaffirmed by each President, complicates the calculations of any aggressor considering attacking France's interests and thereby discourages any attempts to circumvent deterrence. 1.3 The unique nature of nuclear weapons – unacceptable damage 172. The nature, extraordinary power, speed and effects of nuclear weapons make them unique and incomparable to any conventional weapon. It is these intrinsic characteristics that enable them to inflict instant and unacceptable damage. This capability imposes a level of risk that far outweighs any potential gains an aggressor might hope to achieve. In this contest of wills, it is that prospect which must deter any inclination towards aggression. Any use of nuclear weapons in a conflict would irreversibly and fundamentally alter its nature. This is why nuclear deterrence is unique and fundamentally different from any form of discouragement undertaken by conventional, political or economic means. 1.4 The nuclear warning 173. If a state adversary were to take the risk of attacking France, having misjudged the nature of its vital interests, a so-called nuclear warning strike could be carried out against them. Such a strike would be intended to send a clear message to the aggressor that the nature of the conflict has fundamentally changed and to restore the logic of deterrence by compelling them to abandon their ongoing acts of aggression against our country. 174. This nuclear warning is optional, singular and non-repeatable. It is not intended to achieve military dominance over an adversary, as the French nuclear weapon is in no way a battlefield weapon. 1.5 Strict sufficiency 175. France rejects any arms race. The level of French nuclear forces does not depend on the offensive capabilities, whe…

Say what you will about Macron's approach to Europe but the French philosophy on nuclear deterrence kicks ass

📑 www.sgdsn.gouv.fr/files/files/...

The future of AI governance may hinge on our ability to develop trusted and effective ways to make credible claims about AI systems. This new report expands our understanding of the verification challenge and maps out compelling areas for further work. ⬇️

I've normally seen this kind of thing coded as one of the non-extinction outcomes that are possible within the broader existential risk definition that I think Bostrom introduced. For example, AIs enforcing ideological homogeneity.

I can only speak by reference to some of his written work. What I've read so far has been worth the read.

Epoch AI Epoch AI is a research institute investigating key trends and questions that will shape the trajectory and governance of Artificial Intelligence.

You might be interested in work from Anton Korinek and his coauthors. I also recommend checking out Epoch epoch.ai And if you're going to have accelerationist stuff you might want to include Vitalik Buterin on defensive accelerationism: vitalik.eth.limo/general/2023...

If you like the steel drum, check out the handpan too if you haven't already.

16/ Guillem Bas, @nickacaputo.bsky.social, Julia C Morse, Janvi Ahuja, Isabella Duan, Janet Egan, Ben Bucknall, @briannarosen.bsky.social Renan Araujo, Vincent Boulanin, Ranjit Lall @fbarez.bsky.social, Sanaa Alvira, Corin Katzke, Ahmad Atamli, Amro Awad /end🧵

15/ Thanks to @aigioxfordmartin.bsky.social ‬ for backing this project and all my coauthors: Robert Trager, @ankareuel.bsky.social, @davidmanheim.alter.org.il,‬ @milesbrundage.bsky.social, Onni Aarne, @aaronscher.bsky.social, Yanliang Pan, Jenny Xiao, Kristy Loke, Sumaya Nur Adan

Verification for International AI Governance - Oxford Martin AIGI The growing impacts of artificial intelligence (AI) are spurring states to consider international agreements that could help manage this rapidly evolving technology. The political feasibility of such ...

14/ The full report is available here: aigi.ox.ac.uk/publications...

13/ Those who lived through or studied the Cold War may remember President Reagan reiterating the Russian proverb “Trust, but verify.” Just as it was with 1980s nuclear arms control, our ability to build new verification systems may be crucial for preserving peace today.

12/ If we build these more serious verification systems, we would be laying the foundation for international agreements over AI—which might end up being the most important international deals in the history of humanity.

A schematic summary of the concept of verifiable confidential computing. A Prover is able to demonstrate something to the Verifier without either of them revealing to the other the entirety of their secrets (e.g., AI model for the Prover and sensitive test content for the Verifier). They arrange for cryptographic commitments to be made about the activities of the Prover. These can then later be used to ensure that the same data and code is provided in a neutral datacenter where verification computations can take place. These verification computations can be privacy-preserving and thus protect everyone's data while demonstrating that these digital objects follow rules.

11/ It seems possible to create similar verification exchanges that preserve security to an extreme degree, but we’ll need political action to get there. Our report goes into this in some detail. These setups might take about 1-3 years of intense effort to research and build.

10/ However, even if we scale this up, the most important secrets (think national security info, military AI models, or the Coca-Cola formula) are probably too sensitive to govern via just confidential computing. Further work is needed to safeguard these.

9/ Groups that use AI (including corporations and countries) will likewise place more trust in AI services that they can be sure are secure and appropriately governed. They may also request—or demand—this kind of thing in the future.

8/ This setup allows 1) users to feel safe and confident about services they pay for, 2) companies to expand their offerings to more sensitive domains, and 3) governments to check that rules are followed.

7/ An AI provider can prove that they abide by rules by having a set of third parties (e.g., AI testing companies and AI Safety / Security Institutes) securely test their models and systems. A user can trust a group of third parties a *lot* more than they trust the AI provider.

6/ Confidential computing might be reliable enough for a company to make pretty strong claims about what they are *doing* (e.g., serving you inference with a given model and compute budget) and what they are *not doing* (e.g., copying your data).

5/ Some of these technologies can be deployed *today*, such as confidential computing, which is available in recent hardware such as NVIDIA’s Hopper or Blackwell chips. These are good enough to get us started.

4/ Luckily, decades of work has gone into privacy-preserving computational methods. Basically they are tricks with hardware and cryptography that allow one actor (Prover) to prove to another actor (Verifier) something without revealing all the underlying data.

3/ But countries care about their security, so we can’t expect them to simply hand over all the information needed to prove that they’re following governance rules.

2/ International AI governance is desirable (for peace, security, and good lives), but it faces verification challenges because there’s no easy way to understand what someone else is doing on their computer without violating their security.

Cover of the report "Verification for International AI Governance" published by the Oxford Martin AI Governance Initiative. Authors are: Ben Harack, Robert F. Trager, Anka Reuel, David Manheim, Miles Brundage, Onni Aarne, Aaron Scher, Yanliang Pan, Jenny Xiao, Kristy Loke, Sumaya Nur Adan, Guillem Bas, Nicholas A. Caputo, Julia C. Morse, Janvi Ahuja, Isabella Duan, Janet Egan, Ben Bucknall, Brianna Rosen, Renan Araujo, Vincent Boulanin, Ranjit Lall, Fazl Barez, Sanaa Alvira, Corin Katzke, Ahmad Atamli, Amro Awad

Governing AI requires international agreements, but cooperation can be risky if there’s no basis for trust.

Our new report looks at how to verify compliance with AI agreements without sacrificing national security.

This is neither impossible nor trivial.🧵

1/

Fascinating and important.

Cosmos by Carl Sagan was going to be my answer, but now I realize Star Trek should be in there too!

Always a challenging needle to thread, since you want them to be successful (to continue to exist) without being overrun!

The conditions that have led to what’s happening in the US today exist in democracies around the world.
They are an inevitable outcome of our collective failure to adapt to fundamental changes in the information ecosystem on which our democracies were originally built.

🔊 Listen Now: A whistleblower's disclosure details how DOGE may have taken sensitive labor data All Things Considered on NPR One | 7:13

Happy Friday everyone. Thanks for reading NPR.org this week.

Wanted to take a second to also remind you: I interviewed whistleblower Dan Berulis to accompany my lengthy written story on NLRB. Hear from him in his own words:

one.npr.org/i/nx-s1-5355...