Artificial Intelligence

Security Council meets on losing control of AI, briefed by the men who build it

This afternoon in New York, under the French presidency, the Security Council holds its first meeting devoted to the possibility that frontier AI systems slip out of human control. The briefers are Yoshua Bengio and the chief executives of OpenAI, Anthropic and Hugging Face. Whether the Council is the right room is itself in dispute.

Norbert Wiener

Assisted thinking in the Alps, and the rest of philosophy's late September

The 29th Philosophicum Lech opens in Vorarlberg under the title 'Betreutes Denken', a pun on assisted living that puts Kant's word for immaturity next to artificial intelligence. Vancouver's World Congress on Logic and Religion and Penn's Foucault centenary follow within days, and four meetings in Ghent, Lublin, Bucharest and Berlin have just closed.

Immanuel Kant, Michel Foucault

The AI art trial that did not happen, and the question it still poses

Reports that Andersen v. Stability AI went to a San Francisco jury this month are false: the docket shows a case still in discovery, no class certified and trial set for June 2027. Beside Thaler v. Perlmutter, closed by the Supreme Court in March, it shows how poorly copyright's two words, copy and author, fit what artists say was taken.

What AI art exposes is how little we ever agreed about art

Generative systems have not answered the question of whether a machine can be an artist. They have made it impossible to leave the question vague. A survey of what Danto, Benjamin, Barthes, Collingwood, Dewey, Goodman, Adorno, Heidegger and Boden actually help with, and where the courts have got ahead of the philosophers.

Martin Heidegger, Arthur Danto, Walter Benjamin, Margaret Boden

AI and the public sphere: who authors the decisions made in your name

The argument about artificial intelligence and democracy is usually run as a question of benefit and harm. The harder question is authorship. Arendt, Habermas, Foucault, Rousseau, Marx and Norbert Wiener each supply a piece of it, and the 2026 research supplies some awkward facts.

Jürgen Habermas, Hannah Arendt, Jean-Jacques Rousseau, Norbert Wiener

Compute, chips and the responsibility gap: artificial intelligence as geopolitics

The contest over artificial intelligence is a contest over fabrication capacity, electricity and refined minerals. Hobbes explains why states cannot trust one another about it; Kant explains why the answer was never a world government. Both run out at the same place: a killing with no author.

Thomas Hobbes, Immanuel Kant

Four Fields Medals, one machine proof and the question of what the prize is for

Yu Deng, John Pardon, Jacob Tsimerman and Hong Wang received the 2026 Fields Medals in Philadelphia on 23 July. Seven weeks later an AI swarm settled a version of Navier–Stokes and Tristan Buckmaster said the mathematician who built the method deserves a medal. A prize for promise now has to say whose promise it means.

The proof no one has read: OpenAI, Navier–Stokes and the limits of verification

OpenAI says an internal system proved that a three-dimensional fluid can blow up in finite time, and Lean agrees. The theorem is narrower than the headlines, the credit is contested, and the deeper question is what mathematics loses when a result is certified but unread.

John Searle

Is talking to a chatbot communication? An Oxford volume takes the question apart

Herman Cappelen and Rachel Sterken's open-access collection asks whether what passes between a person and a language model is communication at all. Sixteen chapters, one of which argues the volume's own premise is false. A review of the arguments, and of what the book declines to notice.

Hans-Georg Gadamer, Saul Kripke, John Searle