Trump announced a two-week ceasefire with Iran just before his own deadline, but the consensus from analysts is that Iran came out ahead: former US diplomat Aaron David Miller put it bluntly as 'Iran has won another round,' while political scientist Robert Pape writes it's 'the biggest loss since Vietnam.' The deal reportedly uses Iran's 10-point framework over the US's 15-point proposal, with Iran reportedly committing to reopen the Strait of Hormuz—though Trump administration insiders told Zeteo they expect him to ramp escalation back up within weeks. Expert Trita Parsi correctly predicted much of this outcome, while panelist Harrison Mann cautions the Lebanon extension could actually accelerate ethnic cleansing in Gaza by freeing up troops for that front.

The Iran deal's fragility also casts a shadow on Trump's executive power expansion at home — internal FBI emails reveal the administration is directing state and local police to enforce NSPM-7, a directive that criminalizes antifascism, anti-capitalism, and criticism of US foreign policy as domestic terrorism, funneled through over 75 fusion centers nationwide. AG Pam Bondi's December memo routes suspected left-wing cases to Joint Terrorism Task Forces, where dozens of investigations are already active — though cities like Portland and San Francisco have opted out of the fusion center partnerships entirely, a model Boguslaw says other communities can replicate. Meanwhile, Congress has largely surrendered on war powers, and courts' nationwide injunctions remain the last meaningful institutional check — not dead yet, per SCOTUS.

On the AI and security front, Anthropic launched Project Glasswing with 45+ partners including Apple and Google to test a new Claude Mythos model for cybersecurity specifically — a restricted release, because an Anthropic employee noted the model's hacking capabilities are genuinely alarming. That caution looks well-timed given that Iran-linked hackers are already targeting US energy and water infrastructure, and Meta paused work with data vendor Mercor after a breach exposed AI training secrets. Meta's broader AI play, meanwhile, rests less on model quality and more on a distribution moat of 3.5 billion daily users — though chief AI scientist Yann LeCun quit and raised $1.03B for a rival lab pursuing a fundamentally different approach.

Elsewhere: the art world got some marquee news — Kengo Kuma beat out Renzo Piano and Selldorf to design London's National Gallery's £350M new wing, MoMA opened a Duchamp retrospective examining how he redefined art itself, and Spain's culture minister blocked Guernica traveling to Bilbao on conservation grounds. A new ARTnews study found Parisian galleries are still heavily tilted toward Western artists — just 4.7% from Africa, 5.3% from Asia — even as female representation has improved. And on a different kind of rights issue, a Queens workshop is teaching trans people to 'self-dox' and scrub their data from the internet, a necessary precaution as a Kansas law invalidated hundreds of trans IDs overnight and Idaho's felony bathroom bill ratchets up the stakes of being visible.