THE MIDDLE EAST

A peace plan for Gaza, and the first real hope of an end

Two years to the week since October 7, an American plan to end the war gained sudden momentum. Hamas agreed to key terms, Israel signalled acceptance, and for the first time in the conflict's grinding history, the end of it moved from fantasy toward possibility.

Two years after the war began, the first genuine hope of ending it emerged this fortnight. President Trump unveiled a 20-point plan on September 29 to end the war in Gaza, providing for the release of all remaining hostages, a phased Israeli withdrawal, the disarming of Hamas, and a transitional international administration of the shattered territory, and, remarkably, the plan gained real traction where every previous effort had collapsed. Hamas, battered and increasingly isolated, agreed on October 3 to key elements, including the release of the hostages; Israel, under an American pressure it could not easily resist, signalled its acceptance; and the Arab states lined up behind it. Enormous obstacles remained, the disarmament, the governance, the sequencing, the deep mistrust, and the plan could still collapse as others had. But for the first time in two years, the end of the war was a live possibility rather than a distant fantasy, and in Gaza and Israel alike, exhausted people allowed themselves a fragile, disbelieving hope.

The plan's momentum came, ironically, from the same transactional, coercive presidency this paper has spent the year criticising. The dealmaker's instinct, the willingness to pressure friend and foe alike, the indifference to the niceties that had constrained previous mediators, produced movement where principled patience had produced only stalemate. It was an uncomfortable lesson, and an honest paper must acknowledge it: that the same qualities which make this administration so dangerous to alliances and institutions may, occasionally, in the specific arena of a deadlocked war between exhausted parties, produce a result that the more careful approach could not. Whether the plan holds, whether the hostages come home and the guns stay silent and the terrible work of rebuilding begins, is unknown and far from certain. But the possibility itself, after two years of unrelieved horror, was the most hopeful thing this paper has had to report from the region since the war began, and it deserved, cautiously, to be marked as such.

WASHINGTON

The government shuts down, and dysfunction becomes the baseline

The United States government shut down on October 1 as Congress failed to agree on funding, sending hundreds of thousands of federal workers home without pay and suspending a swathe of government services, in a now-familiar ritual of dysfunction that has become so routine it barely shocks. The shutdown, driven by a standoff over healthcare subsidies and spending, was distinguished by the administration's threats to use it as cover for permanent cuts and mass firings, turning a temporary budget impasse into another front in its campaign to dismantle the federal government. The dysfunction that once represented a crisis has become the baseline, a permanent feature of a governing system that can no longer reliably perform its most basic function, keeping itself funded and open.

IN BRIEF

A flood of synthetic video; troops toward the cities; the Nobel season

OpenAI released a powerful new video-generation tool, Sora 2, and an app to go with it, unleashing a flood of hyper-realistic synthetic video, including convincing fabrications of real and dead people, that raised anew, and more urgently, the alarm about a world in which video can no longer be trusted as evidence of anything. The administration moved to deploy troops toward American cities including Portland and Chicago, over the objections of their leaders, extending its pattern of domestic military deployment, and designated its loosely-defined domestic opponents as terrorist threats. And the Nobel Prize season opened, its announcements a reminder of the patient, cooperative, truth-seeking human enterprise that the fortnight's other news so relentlessly assaulted.

The Column

The end of the evidentiary image

OpenAI released a tool this fortnight that generates convincing video of anything, including real people, from a sentence. The worry about fakes is real but too narrow. The deeper event is the end of a brief historical age in which the image counted as evidence.

By Daniel Reeve

OpenAI released a tool this fortnight that generates convincing video of essentially anything, real people included, from a written sentence, and the immediate alarm is about fakes: fabricated footage, invented events, the coming flood of convincing lies. That worry is real, but it is too narrow, because it treats the problem as an increase in fakery to be countered with better detection, when the deeper event is something more fundamental and irreversible: the end of a brief and historically unusual age in which the photographic image counted, almost automatically, as evidence.

It helps to see how strange that age was, and the philosopher Walter Benjamin, writing in the 1930s about what mechanical reproduction did to art, gives the frame. Benjamin noted that the photograph and the film introduced something genuinely new into human experience: a mechanically produced image that seemed to bear an objective, evidentiary relationship to reality, that showed what was actually there in a way a painting never could, because it was made by a machine registering light rather than by a hand that could invent. This gave the photographic image an authority no previous image had possessed, the authority of evidence, of witness, and for roughly a century and a half that authority was so reliable that it reshaped law, journalism, memory, and the very phrase we use for certainty: I saw it with my own eyes, seeing is believing. But this evidentiary status of the image, we are now realising, was not a permanent feature of images; it was a temporary artifact of a particular technological moment, the moment when making a convincing image required capturing something real, and that moment is now ending, because a machine can now generate a perfectly convincing image of something that never happened, severing the link between the image and the real that gave the image its authority in the first place.

The counterargument from the optimists is that this is manageable, that we will develop detection tools, provenance standards, cryptographic authentication, and learn to verify images as we once learned to verify documents, so that the evidentiary image survives in a new, authenticated form. There is real hope here, and the technical efforts are serious and worth pursuing. But the optimists understate how deep the change goes, because the evidentiary power of the image was never really in the image; it was in the fact that a convincing image was expensive to fake, that the default assumption could be trust because fakery was hard, and once fakery becomes trivial the default has to flip from trust to suspicion, and no amount of authentication technology restores the old condition in which you could simply believe your eyes. Authentication can tell you that a particular image is verified; it cannot restore a world in which the unverified image carries any weight at all, and most images, most of what floods past us, will never be authenticated. We are not adding fakes to a world of trustworthy images; we are ending the world in which an image, as such, could be trusted.

And that returns us, strangely, to where humanity was before the camera, which is the deep lesson worth carrying. For almost all of human history, before the photograph, truth about distant events rested not on images but on testimony, on the credibility of witnesses, on chains of trust in the people and institutions that reported the world to you, and the evidentiary image was a century-and-a-half interruption in that older condition, a period in which the machine-made picture let us bypass the question of whom to trust and simply see for ourselves. That interruption is ending, and we are being returned to the older world in which seeing is not believing and truth about what we cannot witness directly must once again rest on trust, on provenance, on the credibility of sources rather than the evidence of images. This is not the end of truth, as the panic suggests; humanity knew truth before it had trustworthy images and can again. But it is the end of the shortcut, the end of the age when the image did the work of trust, and it means that the institutions of credibility, the sources and witnesses and verifiers we choose to believe, matter more now than they have in a hundred and fifty years, because the image can no longer stand in for them. The flood of the fake is real. But the thing it washes away is not truth; it is the brief and comfortable age in which we did not have to ask whom to trust, because we could just look.

The Ledger

Sora 2's flood of convincing synthetic video crosses the threshold this paper has warned of for two years: the end of video as trustworthy evidence, and thus the end of "seeing is believing." The crisis of truth deepens from a problem of institutions to a problem of evidence itself.

Data centers & power

The generative-video boom means yet more compute, yet more power; every synthetic video is energy spent, and the flood of AI "slop" now filling the internet is, among other things, a vast and growing draw on the strained grid.

Rates

The government shutdown adds fiscal dysfunction to the mix, though markets, inured to the ritual, largely shrugged. The Fed's easing continues against a backdrop of political chaos its cuts cannot cure.

Real estate

US 30-year mortgage 6.30 percent (October 2), near the year's low as the economy cools and cuts continue. The shutdown's economic drag, if prolonged, could push rates lower still.

India tech

The synthetic-media flood and the AI-governance retreat raise the stakes for India's technology sector: the firms that can build trustworthy, verifiable, auditable systems, the antidote to the flood of fakery, have an opportunity, if they seize it, to be the grown-ups the age desperately needs.

What we called wrong

Nothing to retract. Sora 2 confirms, on schedule, the warning this paper first made as this war began two years ago: that the age of synthetic media would end the trustworthiness of evidence, and that a society would have to relearn how to know things without the certainty of its own eyes.

The Back Page

Two years

It is two years, this fortnight, since the morning of October 7, and the anniversary fell in a week when, for the first time, the end of the war it began seemed genuinely possible. It is worth, on the anniversary, holding the whole terrible arc in view. Two years ago Hamas killed 1,200 people in a single day and took 250 hostages into the tunnels of Gaza. In the two years since, the war that followed has killed tens of thousands of Palestinians, reduced much of Gaza to rubble, brought famine to its people, drawn in Lebanon and Iran and America, and held the world in a grip of grief and rage and argument that has not loosened for 730 days. Some of the hostages came home, in the ceasefire in January and in deals since; some were killed in captivity; a remnant remains held, two years on, their families still waiting. The peace plan on this fortnight's front page may, if it holds, finally bring the last of them home and finally silence the guns. Or it may collapse, as every previous hope has, and the killing may grind on. But on the anniversary of the day it started, it is worth remembering what all of this has cost, on every side: the Israeli families destroyed on October 7 and the hostages still in the dark, the Palestinian families destroyed in the two years since and the survivors starving in the ruins, the whole vast human catastrophe that two years of war have piled up, and that no plan, however hopeful, can undo. If the war ends now, it will end having taught, again, the oldest and least-learned lesson: that the descent into catastrophe is swift and the climb out of it is agonisingly slow, and that everything the two years destroyed will take far longer than two years, if it is ever possible at all, to rebuild.