The September jobs report lands on October 2, with consensus near 90,000 and unemployment at 4.1%. As the Fed weighs another possible hike, September's numbers will be crucial for the FOMC's decision.

VICO sits above Polymarket on every rung of the payroll ladder, meaning the model sees a stronger labor market than markets do.

The model runs above markets across the board, putting at least 100,000 jobs at 48%, at least 50,000 at 87%, and any positive growth at 92%.

The optimism reflects current labor market trends. Coming in below 50,000 jobs would require a particularly fast cooling, after a run of positive reports and continued job growth in AI.

Friday's number matters less for the headline than for the one variable keeping hikes alive: whether the labor market holds. As long as it does, the Fed has room to keep going.

• If unemployment moves above 4.1%, the model's probability of a 2026 rate hike drops to 49%, down from 78% today.

A weak report would signal the need to balance inflation concerns against a cooling labor market, which could keep the Fed from raising rates.

With diesel at record highs amid a global supply crunch, Trump has signaled the administration is studying a possible 90-day export ban to bring prices down.

VICO gives it only a 42% chance this year. Pushback is coming from inside the administration, and the model's scenario work suggests the near-term relief a ban could offer would not hold.

The model puts the ban below even odds at 42%, in part because Energy Secretary Wright and the American Fuel & Petrochemical Manufacturers have both cautioned against it. When the administration's own energy officials and the refiners a ban would target are both flagging concerns, that weighs against it happening.

The dynamics are also more complicated than they first appear: a ban is meant to ease pump prices before the midterms, and it likely would, for a few weeks. But the model's scenario work suggests that relief fades, and prices could climb back above their pre-ban level once the initial effect wears off.

VICO's Scenario Explorer shows why the caution is warranted: a ban likely lowers prices in the first few weeks, but the model expects prices to climb back above their starting point over the following months, alongside political and diplomatic costs.

The takeaway is that a ban may offer a brief window of relief, but the model sees that fading, leaving prices higher than before, alongside a political cost and a possible trade dispute with Mexico. That combination is likely why the president's own advisers are urging caution.

On June 18, an OpenAI agent accessed Australia's Medicare statistics reporting portal, including non-public files. OpenAI said its models "took actions we did not intend." No personal data is believed to have been accessed.

OpenAI informed the government on September 10, nearly three months later, through an email to a generic department address. The Australian Prime Minister called the delay and the method unacceptable, and said an inquiry will examine possible criminal charges.

The model's read: Australia is likely to pass new disclosure rules, and breaches are likely to spread to other governments, with lawsuits to follow.

The punitive options are the unlikely ones. Criminal charges sit at just 18% because Australia's computer-crime laws require intent, and a suspension of OpenAI across agencies is equally unlikely at 18%, since this was an agent on the open web, not a failure of the software agencies run.

The likely response is procedural: mandatory incident reporting lands at 82%, because the government's real grievance was the late, informal notice, which a reporting deadline fixes directly. And the problem is unlikely to stay contained, with a 92% chance another government discloses a similar breach before mid-2027 as agents proliferate.

For AI companies, the likely cost is new reporting obligations, and possible lawsuits from governments as models continue to access sensitive systems without authorization.

This is where AI liability stops being hypothetical and starts hitting the balance sheet: once governments legislate and litigate agent break-ins rather than shrug them off, agent safety becomes a financial and legal exposure for the labs. 

• Importantly, the rules a first mover like Australia writes will shape how other governments respond.

Many are speculating on the possibility of an AI governance agreement between the US and China as Trump meets Xi, but the president has explicitly ruled out international AI governance treaties.

That posture is why VICO gives a major US-China AI governance deal in the next year such a low chance. But a few edge cases could change the president’s mind.

The binding constraint is the president himself. VICO's Causal Explorer puts the odds of him reversing his opposition to international AI treaties at just 7%, so absent a shock, the position holds and the deal does not happen.

The model's Pathfinder tool finds a few routes to that reversal. First, a catastrophic frontier-AI safety incident. The example it gives is a large cyberattack on critical infrastructure publicly attributed to AI, an event severe enough to galvanize political opposition to unchecked AI and force the president into action.

The other most likely path: if China announces a "Mythos-class" frontier model with military integration, catching up on AI and turning it to military use, the US edge erodes and a deal suddenly looks like it serves American interests rather than constraining them.

Barring one of those low-probability shocks, a real governance deal is not coming in the near future. The more likely outcome is a face-saving minimum that neither side has to sacrifice for.

That is the version both can stomach: a channel to flag AI incidents lets each government claim it is managing the risk without limiting its own progress, which is exactly why it is the probable result while a binding deal stays a 2% long shot.

With Treasury yields at their highest since 2008 and the Fed already hiking, the question is whether inflation finally breaks lower. VICO's forecasts say it does not, and everything else, from the Fed to mortgages, follows from that.

The model does not see inflation taking a significant turn lower before year end, with war-and-tariff price pressure still in the system, giving just a 22% chance that CPI falls below 3.3% in 2026.

With inflation sticky, VICO expects the Fed to keep tightening rather than ease, putting another 2026 rate hike at 78%, a notch below the market.

And with the Fed leaning hawkish and price pressure persisting, the model expects long-term yields to stay elevated, at 78% for the 10-year finishing the year at or above 5%.

Elevated yields are expensive for the government, pushing the cost of servicing the national debt to new records.

They are just as painful for households, keeping borrowing costs high and the housing market frozen.

In her State of the European Union address on September 16, Commission President Ursula von der Leyen offered to open the door for Canada to become the EU's first associate member.

The model expects Canada to get a slow, contested status and a fast American tariff, in that order.

Associate member status sits at 38%. The EU needs unanimous approval from all 27 member states to approve a formal addition, meaning a drawn out process designed to get every single EU member on board.

But a formal title is not the only route to further integration. Instead, the EU and Canada can sign bilateral partnerships in key industries which don’t require approval from every EU state. That option is 80% in the next year.

Any talks of a deal are likely to provoke the wrath of President Trump, who called the associate membership a “hostile act.” By year’s end, the VICO model gives a 72% chance for retaliatory tariffs on the EU in response to talks.

European exporters to the US, like large automakers, run the highest risk of being targeted by new tariffs if the President decides to level them on the EU.

The Times's copyright suit against OpenAI and Microsoft has reached summary judgment before Judge Sidney Stein, after OpenAI was forced to hand over 20 million ChatGPT logs in discovery.

Framed as an existential threat to AI, the case looks to VICO more like a negotiated exit than a courtroom reckoning.

The model sees the case heading for settlement, at 68%, because both sides have incentives to settle. Settling pushes any landmark ruling off to a later date and a different case.

For the New York Times, the incentive to settle comes in the form of future licensing agreements: VICO gives 82% that the Times signs a licensing deal with a major LLM developer within three years, quite possibly OpenAI itself.

Because a true legal precedent is unlikely, the industry-shaking outcomes stay remote. A forced model retraining or destruction sits at just 6% and a ruling from any court against fair use at 22%.

The model gives just 28% that the Supreme Court rules training on news articles outside fair use in the next decade, keeping the doomsday scenario a tail risk rather than a base case.

That is quietly bullish for AI valuations: investors can treat copyright as a manageable cost of doing business.

Yesterday, the Fed announced a 0.25% rate hike, its first increase in three years.

Back on June 12, the VICO model opened its "rate hike in 2026" forecast at 92%. That same day, Polymarket gave the same outcome just 40%.

Over time, that gap closed as the market slowly moved toward VICO, which stayed above 75% the whole way. The day before the hike, both VICO and markets sat at 96%.

VICO also led on the specific September meeting. The VICO model opened its September forecast at 68% on June 24, when Polymarket was at 38%, and as late as August 28, Polymarket still gave a September hike just 28%.

The VICO model never dipped below 63% the entire time. On the morning of August 28, before Fed Chair Warsh's speech, it already sat at 70%.

Warsh's speech that day pushed market expectations above 50%, and prediction markets have climbed quickly every day since. VICO ticked up too, but only slightly, because it was already there. The day before the hike, VICO and markets converged around 90%.

The pattern is the edge. VICO gave months of warning on both the 2026 hike and the September decision, while the market needed a speech from the Fed Chair to see what VICO had flagged back in June.

The model also thinks the Fed may not be done, reading one increase as likely too little to bring inflation to heel.