FutureOfCitizenship

“AI isn’t an independent speaker...It’s a tool that people use to write, research and communicate ideas.” - John Coleman of FIRE, a free speech non-profit, on Minnesota bill SF 4114. Can chatbots speak? And if they can, do they have free speech, like humans? See our forthcoming paper: papers.ssrn.com/sol3/papers.cf… @jeffrsebo @JonSimonSays

@AdamThierer It's an automatic, rotating presidency. The U.N. is run by member states. It has no power to "appoint" any state to do anything.

some AI safety advocates call for int'l regulatory controls, possibly coordinated through the United Nations. Meanwhile, the U.N. appoints Iran and North Korea​ to leadership positions on nuclear non-proliferation issues. ​😵 Why would we ever trust the U.N. with anything??

This seems like a strange way to summarize a book that ends with its author sitting in a cave by himself, no closer to his goal of understanding consciousness than when he started. theatlantic.com/books/2026/02/…

Interesting paper on AI rights from some of my fellow human rights lawyers, and so glad to see more HR lawyers engaging with the topic. Something to note: we lawyers must engage more with the lit on AI moral status and consciousness, since it goes to the heart of the question. A summary like the one in this paper is simply not enough. The last time we granted rights and duties to an autonomous entity based on anything other than moral status was in the Middle Ages; doing so is less of a "pathway" and more of a revolution in the very idea of rights. Also, there's not enough in here of how we would hold AI systems to account if they fail to fulfill their duties. papers.ssrn.com/sol3/papers.cf…

Arbel and Tokson talk AI risk in their forthcoming paper: "A more nuanced view recognizes that transformative technologies rarely develop as either proponents or critics initially expect. The internet was simultaneously overhyped in many ways (it did not bring about global democracy) and underhyped in others (social media transformed culture, politics, and the economy). Nuclear technology followed a similar pattern: initial euphoria about peaceful applications gave way to sobering recognition of proliferation risks, while early fears of widespread nuclear conflict proved overstated—though, crucially, only because of deliberate policy interventions including arms control treaties and nonproliferation regimes." papers.ssrn.com/sol3/papers.cf…

SCOOP: The White House is developing guidance that would allow agencies to get around Anthropic's supply chain risk designation and onboard new models including its most powerful yet, Mythos. axios.com/2026/04/29/tru…

Google abruptly dropped out of a $100 million Pentagon prize challenge to create technology for voice-controlled, autonomous drone swarms after it was among the successful submissions, according to people briefed on the matter. bloomberg.com/news/articles/…

The NY Times has set up live commentary for Musk v. Altman. Free article. nytimes.com/live/2026/04/2…

Galileo and the Movement of the Spheres. I review A World Appears, by Michael Pollan.

The ITU is also super important for global AI standards.

Glad to see something gets positive valence from reviewing legal contracts.

So this is consistent with the Pentagon’s January strategy memo which requires AI systems for “any lawful use” and “free from usage policy constraints that may limit lawful military applications.” This makes clear that the “safety stack” isn’t going to stop anything that the government deems to be lawful. Any dispute over whether something is lawful will be after the fact if the agreement contains a typical government disputes clause. No one should be surprised by these terms. “Any lawful use” means “any lawful use.” I explain this in greater detail here (but with respect to OpenAI, not Google). lawfaremedia.org/article/milita…

arxiv.org/abs/2405.07987 Do all roads lead to the Cave?

A new paper from Google Deepmind explaining why just watching a video of a shower every morning hasn't been working out for me. philarchive.org/rec/LERTAF

See our substack: "Understanding Anthropic v. the Pentagon: Is the solution for the USA to build its own AI models?" at futureofcitizenship.substack.com/p/a-reading-li…

wsj.com/politics/polic… (gift) A lot going on in the regulatory space on AI safety in the USA at the moment.

(Gift article) White House pushed out new AI official after just four days on the job wapo.st/3OQ7YOB

This is a great paper from researchers at Microsoft on why consciousness matters and the dangers of false attribution. But it can also be dangerous to present rights as a zero-sum game: "Moral patienthood judgments carry material implications. If (in-group) are misclassified as deserving moral consideration, resources may be directed towards them...As resources are finite, such diversions represent zero-sum reallocations away from (out-group)...An erroneous allocation of such rights based on (undefined or disputed classifier) would lead to significant and unwarranted societal disruption." papers.ssrn.com/sol3/papers.cf…

"But courts have yet to decide how much human involvement is required." Judges have been dancing around this copyright question for years now. Some judge needs to go first. Anthropic’s Leaked Code Tests Copyright Challenges in A.I. Era nytimes.com/2026/04/22/tec…