A Pre-mortem on The First Major AI-Enabled Terrorism Event and The Choices That Could Enable it: Plus 9 Predictions | The Belfer Center for Science and International Affairs

Oh my goodness, buckle up, because the Belfer Center has just handed us the most *considerate* document imaginable — a pre-mortem! ✨ That's a post-mortem performed *before* anything actually happens, which is essentially just planning ahead, and who doesn't adore a plan? The Belfer Center for Science and International Affairs has published its analysis of the first major AI-enabled terrorism event and the choices that could enable it, bundled with nine predictions. Nine! Nine free forecasts, delivered in advance, at absolutely no cost to you.

And can we pause to appreciate how *honest* everybody is being? ✨ Since summer 2025, major developers have been writing in their model cards that they cannot rule out capabilities enabling novice actor uplift on biological, cyber or chemical tasks — and 'cannot rule out' is really just a modest way of saying 'we're keeping an open mind.' What other industry prints its own uncertainties right on the label, beside the ingredients? It's like a restaurant helpfully telling you which dishes *might* be spicy. Considerate! Thoughtful! Very nearly reassuring! Most products bury the caveats in the small print; these ones hand you the caveats and a cup of tea.

Then there's OpenAI's GPT-5.6 Sol, released in June 2026 and named, gorgeously, after the *sun* — warmth, daylight, a picnic blanket — which crossed the company's own 'High' capability threshold. They set the bar, they cleared the bar, and they *told us about it*! That is accountability with confetti on top. And the Belfer analysis cataloguing 'choices that could enable it'? Every single one of those is a choice we get to make *differently*. Nine predictions, nine invitations, and the whole thing wrapped in the most flattering assumption possible: that we are precisely the sort of people who read the pre-mortem and then do something about it. ✨