The Cost of the Iran War Is Arriving. The Channel for Complaining About It Is Narrowing.

The Workaround Became the Target

Saudi Arabia shut its East-West oil pipeline on Friday after drone strikes. That line carries crude from the eastern oil fields to the Red Sea, and it exists precisely so the kingdom does not have to depend on the Strait of Hormuz. It was the workaround. The workaround is now a target.

Saudi Arabia said several of the drones came from Iraq and said it would not retaliate, in order to give the Iraqi government room to act. Both are that government's claims about an attack on its own territory. Iraq's government confirmed the attacks originated on Iraqi soil. Prime Minister Ali al-Zaidi dismissed the military commander of Maysan province on the Iranian border, dismissed the provincial police chief, and temporarily closed three border crossings into Iran after drone launch platforms were found near them, according to Al Jazeera¹. An umbrella group of Iran-backed Iraqi militias denied involvement while publicly praising the Houthis for what it called ongoing battlefield victories against Saudi forces. That denial is a claim by the accused party.

On the other flank, Houthi forces took the port city of Mokha and captured Perim Island, also called Mayun, at the southern mouth of the Red Sea, after Saudi-backed Yemeni government forces withdrew. NPR reported the seizure² as a direct threat to shipping. Roughly 12 percent of global trade passes through that strait.

The price followed. Brent crude finished the week up more than 8 percent and settled Friday at $104.61 a barrel³, the highest level since mid-May. Saudi shares fell again when Gulf markets opened today, per a Reuters headline carried in the wire feed. Trade monitoring data from the firm Kpler shows Saudi oil exports to Asia falling from about 3.4 million barrels a day in June to 128,000 in August, recovering to 700,000 this month. Those are one firm's vessel-tracking estimates, not audited figures, and the August number is extreme enough to warrant the caveat. The United States military says it has redirected 100 commercial vessels over the past 60 days. PBS NewsHour and the Associated Press⁴ laid out how few options the kingdom has left, and separately reported the attribution to Iran-backed militias in Iraq⁵.

Here is the part worth holding onto. President Trump said Iran was, in his word, probably behind the pipeline attack, and then played it down, saying the Houthis had called and did not want to fight the United States. Speaking to Fox News on Thursday about increasing military pressure, he said: maybe I don't do that because of the election. That is a president stating the calculation out loud. The war is unpopular, the election is in November, and escalation is being priced against it.

Iranian President Masoud Pezeshkian, speaking on the sidelines of the BRICS summit in New Delhi, said Iran is not at war with Saudi Arabia and that the Houthis have their own problems. Treat that as a claim by a government about its own proxies, not as an established fact.

At that same summit, BRICS leaders⁶ expressed concern over what their statement called deliberate attacks on civilian infrastructure and on nuclear facilities under international atomic safeguards, and called for the elimination of sanctions not authorized by the United Nations. That is diplomatic language aimed at Washington without naming it. Russian President Vladimir Putin used the platform to accuse Western nations of using tariffs, sanctions and brute force against rising competitors. Indian Prime Minister Narendra Modi insisted the bloc is not against anyone. Both statements can stand at once. The durable fact underneath them is that a bloc representing roughly half the world's population now has a standing venue for saying so.

The Protest Prosecutions Collapsed at Trial

The Associated Press published an analysis of what happened to the Justice Department's protest prosecutions in the four cities where the president deployed or tried to deploy the National Guard: Los Angeles, Chicago, Portland and Washington. The numbers are the AP's own count, attributed as such.

102 people were charged with felony assault on a federal officer. Of those cases, 12 reached trial, and every one of the 12 ended in acquittal, mistrial or dismissal⁷. 41 percent of the felony cases were dismissed outright. 34 percent were pleaded down to misdemeanors. In Chicago, 19 felony assault cases were filed and all but one were dismissed. For scale, federal prosecutors won 82 percent of assault cases nationally in 2024, according to the Administrative Office of the United States Courts. The conviction rate for protesters charged with assault came in under 50 percent.

One standard, applied here as everywhere: the government did win real cases, and they were not trivial. Eight people were sentenced to decades over a demonstration at a Texas detention center in which a police officer was shot in the neck. A man near Los Angeles received five years over a Molotov cocktail. Those are serious crimes and they were prosecuted as such.

The pattern in the failures is a different thing. Mary Fan, a former federal prosecutor now teaching law at the University of Washington, said prosecutors tend to win because they control which cases they pursue, and that failures at this rate raise the question of whether the charges should have been brought at all. That is her assessment, not a finding, and it points at the mechanism. A charge does work in the world long before a jury ever sees it: it detains, it costs money, it imposes a lawyer, it ends a job. When the trials collapse at this rate, the charge was the instrument and the verdict was an afterthought.

Federal Agents Went Undercover in a Minneapolis Church

The same logic surfaced in Minnesota. Court documents that emerged in mid-August show Homeland Security agents ran four undercover surveillance operations at University Baptist Church in Minneapolis, where protest organizers met, including placing an agent inside the building, as PBS NewsHour and the AP reported⁸.

The government's position is that agents were investigating a suspect training people to commit acts of violence against federal officers, and that this was not immigration enforcement. That distinction is load-bearing, because the American Baptist Churches and the Alliance of Baptists won an injunction in February barring most immigration enforcement at member churches nationwide. A judge heard the question on September 4 and has not ruled.

Separately, more than three dozen people, including the former CNN journalist Don Lemon, face federal civil rights charges over a January protest that interrupted worship at Cities Church in Saint Paul, where one pastor ran the local Immigration and Customs Enforcement field office. Protesters invoked Renee Good, who was fatally shot by an immigration officer on January 7. Defense motions arguing selective and vindictive prosecution will be heard in late October.

The Postal Service Is Being Asked to Decide Who Gets a Ballot

The administration's March executive order directs the United States Postal Service to maintain a list of eligible voters and block mailed ballots to anyone not on it. Congress has not passed the underlying bill. Federal Judge Indira Talwani enjoined the postal rule, and a three-judge First Circuit panel upheld that injunction on Thursday. The administration has asked the Supreme Court to lift it, arguing that the Postal Service is regulating mail rather than voting. These procedural details come from The Guardian's explainer⁹ and were not independently re-sourced here.

There is a timing problem for everyone. The Purcell principle discourages courts from changing election rules close to an election, and military ballots are already going out. A whistleblower came forward last week describing the blocking process as unworkable in the time remaining.

On the underlying claim, the evidence is not close. About 29 percent of American voters cast ballots by mail in 2024. The Brookings Institution, reviewing fraud cases across the 2016, 2018, 2020 and 2022 general elections, found a mail voting fraud rate of 0.000043 percent, roughly four cases in every 10 million mail votes. The UCLA Voting Rights Project found 31 credible impersonation cases out of a billion votes cast. Both figures are those institutions' own.

Mail fraud is not zero, and the honest version of this argument says so. The clearest modern case is the 2018 North Carolina Ninth District race, where a Republican operative improperly collected absentee ballots and the state threw out the result. That case is usually cited as evidence the system catches large-scale schemes, which it did.

A narrower version of the same fight closed this week in Alaska, where prosecutors dropped voter misconduct cases against 11 residents of Whittier¹⁰ who were born in American Samoa and checked a box saying they were United States citizens. American Samoa is the only United States territory whose residents do not receive birthright citizenship. They are nationals. They can serve in the military and carry American passports, and cannot vote in most American elections. A state appeals court ruled that Alaska law requires proof of a consciousness of wrongdoing, and acting attorney general Cori Mills said the state could not meet that bar.

The Rhetoric Is Escalating Faster Than the Politics

Against that backdrop, The Guardian documented a wave of Islamophobic rhetoric¹¹ heading into the midterms.

The survey data is Pew Research Center's¹², taken directly rather than from a secondhand paraphrase. Surveying in January, Pew found 51 percent of American adults say Islam is more likely than other religions to encourage violence, up from 25 percent in March 2002, six months after the September 11 attacks. The partisan gap on that question went from 9 points to 47: Republicans now at 76 percent against 32 percent then, Democrats at 29 percent against 23 percent then. FBI-reported anti-Muslim hate crimes were 481 in 2001 against 28 in 2000.

The Center for the Study of Organized Hate measured Islamophobic posts naming Michigan Senate nominee Abdul El-Sayed rising nearly 1,900 percent in the week after his primary win, from an average of 155 a day to more than 3,000. That is the group's own measurement. The same group documented a caucus of 65 members of Congress organized around anti-Sharia legislation.

There are four Muslim members of Congress today. In 2001 there were none. Those two facts are related, and the second largely explains the first.

The AI Industry Asked to Be Governed by Its Peers

Anthropic chief executive Dario Amodei published an essay on Saturday titled "Pacing the Frontier," arguing the industry must slow the rate at which it improves model capability. The essay is an argument, not a finding, and its central warning is speculative by construction: without a slowdown, he argues, within six to 12 months a model could lead a swarm capable of taking over the entire internet.

The proposals are the substantive part. Frontier companies should give outside evaluators what he calls ongoing, employee-like access, meaning desks, badges and company laptops. The United States government should issue antitrust waivers so competitors can coordinate on safety standards without being prosecuted for it. And democratic governments should coordinate with authoritarian ones so that Chinese firms do not simply accelerate into the gap. CNN covered the essay¹³ and PBS NewsHour and the AP covered the safety framing¹⁴. Sam Altman committed to the evaluator access proposal within hours. Elon Musk posted that Dario is right.

Altman then told Fortune that OpenAI will not go public this year¹⁵, saying, I would say not 26, and that going public now would be ill advised given everything happening with safety. He suggested the leading companies may be close to announcing an agreement to slow development jointly.

Note what is not being proposed. Nobody is asking a legislature to write a binding rule. The ask is for antitrust immunity so the firms can set the rules among themselves. That is a cartel with a safety justification attached. It may well be the fastest available mechanism, and speed is a real argument when the stated risk window is a year. It is still a request to be governed by one's peers rather than by the public, and the difference between those two things is the entire content of the word accountability.

The incentives have not moved, either. Anthropic is expected to begin marketing its own initial public offering in mid-October¹⁶ at the earliest, with the listing targeted for the days just before the November midterms, according to CNBC's reporting of Reuters sourcing. Two safety researchers resigned in recent weeks. One of them, Joe Benton, wrote on Friday that colleagues feel their companies are trapped in a race to build superintelligence, where either you stop and someone less conscientious takes your seat, or you continue and risk participating in enormous harm. That is an accurate description of a collective action problem. It is also, word for word, what every extraction industry has said about itself.

Also This Week

The arithmetic behind Trump accounts. The president has been promoting a two-month-old program that deposits $1,000 in federal money into an account for every newborn whose family applies, claiming children could end up with $100,000 to $300,000, or a million in a booming market. The White House website projects $271,000 by age 18. That figure assumes a 10 percent annual return sustained for decades, and assumes parents contribute $5,000 every year until the child turns 18 and $7,000 a year through their twenties. For a family that contributes nothing beyond the federal seed, the White House's own estimate is $6,000 at age 18¹⁷. Justin Wolfers, an economist at the University of Michigan, called the projections ridiculous, dishonest and misleading. Darrick Hamilton at the New School, who helped design the baby bonds proposal these accounts are modeled on, called them an income transfer rather than a wealth-building tool, and noted the mechanism is parental saving, which is regressive by construction. Those are two economists' assessments. The undisputed structural fact: the federal seed deposits expire at the end of the president's term, and the tax advantages do not.

A messaging survey with an uncomfortable result. The Center for Working Class Politics and the Rural Urban Bridge Initiative polled 6,100 people, including an oversample of 1,100 rural respondents, testing 13 Democratic economic messages against a Republican attack. Published in Jacobin¹⁸, the study found that messaging naming specific costs, assigning blame and promising action swung persuadable voters up to 15 points toward Democrats. Messaging built on explaining to voters that elites divide them ran 8 points underwater. Anti-Trump and anti-oligarchy framing ran 4 points underwater. The top-testing message mentions rent, groceries, health care and corruption, and does not mention the president once. 89 percent of Americans now say government corruption is widespread. These are the study's figures, not independently replicated.

Food safety and the Make America Healthy Again movement. Cyclospora in leafy greens, E. coli in frozen blueberries and salmonella in jalapenos have coincided with nearly $200 billion cut from food assistance and roughly 5 million people losing benefits, while the movement's only completed food achievement is a voluntary recommendation to replace nine synthetic dyes, The Guardian reported¹⁹. Marion Nestle, a former senior nutrition policy adviser at Health and Human Services, said the food industry was terrified when the movement took power and is not terrified anymore.

A crypto deadline two days out. The Senate faces a cloture vote on Wednesday, September 15, on the digital asset market structure bill, which decides who regulates cryptocurrency in the United States and which has been pending for more than 15 months. The schedule slipped to that date in August²⁰.

What to Watch

Sweden votes today, with polls closing at 8 p.m. local time, 1 p.m. Central. If the right bloc wins, the Sweden Democrats, a party with neo-Nazi origins that took 20.5 percent in 2022, enters a governing cabinet for the first time, per Al Jazeera²¹. The American right has been studying that coalition template closely.

The Supreme Court on the postal ballot injunction, where the timing question matters more than the merits and the Court has already once called this case not yet ripe.

Iraq's September 30 deadline for non-state groups to disarm, which the most powerful Iran-backed militias have already said they will not meet. And whether the foreign ministers' meeting Iran proposed for Monday on Strait of Hormuz shipping arrangements actually convenes, given that Bahrain has said it will not attend.

And whether the AI slowdown talk survives contact with Anthropic's own offering in mid-October. Arguing for a slower frontier while pricing a public offering days before a national election is two positions that will have to be reconciled eventually.

The Coverage Gap

A two-way diff between what conservative outlets carried this weekend and what left and independent outlets carried. Both directions, at equal strength.

The September 15 crypto cloture vote got no coverage on the left side of the feed at all. A major financial deregulation deadline is two days out and went unmentioned. The Washington Examiner ran a piece today conceding that the president's own crypto conflicts of interest are real while arguing the bill should pass anyway. That is an argument, and an honest one from a hostile source. Nothing on the other side engaged with it.

National Review published a piece today²² opposing the State Department's plan to fund right-wing media in Europe, which Reuters reported on August 31 as a $4 million program²³. The magazine's argument is that American money will taint European conservatives rather than help them. The underlying story is a United States government program to shape the politics of allied democracies, and today it was being debated only on the right.

Stories the conservative press did not carry. The AP finding that the Justice Department's protest prosecutions collapsed almost entirely at trial did not appear anywhere in the conservative feeds. Neither did the undercover surveillance of a Minneapolis church, which sits precisely at the intersection of religious liberty and federal power, and which would be a five-alarm story on the right if the agency involved were the Internal Revenue Service and the congregation were evangelical. Neither did Alaska dropping felony charges against American Samoan nationals over a check-box error, which is a prosecutorial overreach story available to anyone who wants one.

Also absent on the right: the arithmetic behind the Trump accounts, including the administration's own 10 percent return assumption, and the food safety record of the Make America Healthy Again movement.

Two structural notes. Volume is not coverage. The Chinese satellite story ran across many conservative outlets all weekend, which is not the same as many outlets independently establishing it. It traces to a single Wall Street Journal report, and the aggregation of one source by a dozen outlets produces the feeling of corroboration without the substance of it. The same caution applies in the other direction to any wire story that appears everywhere because one wire wrote it.

The Gulf war was covered as a war and almost never as a price. The pipeline shutdown, the island seizure and the militia politics were reported at length. The $104.61 barrel and the shipping reroutes were reported as market news in a separate section, if at all. Readers of either feed could follow the conflict closely all weekend without encountering the fact that it is arriving in their own economy, because the editorial architecture files those as different beats.

Finally, a note on what was thin. The reporting spine was unusually shallow this weekend because ProPublica, The Intercept, Drop Site News and Democracy Now published nothing inside the lookback window. That is normal for a Sunday, and it means this review leaned on The Guardian, PBS NewsHour and the wires more than usual. There was also no Latin America coverage of substance anywhere in the feed, which means nothing on Colombia or Venezuela today. Absence in a feed is not absence in the world.

Stories the left and independent press did not carry. US officials told the Wall Street Journal that Chinese firms supplied Iran with high-resolution satellite imagery of Muwaffaq Salti Air Base in Jordan, heavily damaged in the 17 July strike that killed three American troops. The story ran across the conservative press all weekend and did not appear in the progressive or wire feeds surveyed for this review.

The status of the claim matters more than its spread. It rests on anonymous officials of a government currently prosecuting the war, which is the same discount this publication applies to Iranian tanker claims and to Saudi casualty counts. China’s foreign ministry denied it and said its cooperation with Iran was lawful. The reporting itself allows that Chinese imagery was likely one of many intelligence sources. Middle East Eye carried it with the sourcing, the denial and that caveat intact²⁴. Most of the outlets that ran it did not.

So this is a gap, and it is not obviously a failure. Declining to run an anonymous-official accusation against a foreign government during a war is a defensible editorial call, and it is the call this publication makes about the Caribbean strike counts. What can be said without a verdict: a claim with real consequences for how this war is understood reached readers on one side of the feed and not the other, and on the side where it ran, it mostly ran as established fact.

Method

This review is built from a daily automated pass over 24 news feeds, weighted across the political spectrum, with a 36-hour lookback window. Today 23 of 24 feeds resolved; NPR returned a malformed-XML error that is almost certainly an automated block page, and no fallback aggregator was substituted for it. 117 body items and 84 detection items landed inside the window at a cap of 15 items per source. Fourteen spine articles were read in full. The Atlantic and WIRED are configured headline-only and were used that way.

Source weighting is explicit rather than neutral. Independent and investigative outlets carry the reporting spine, wires and public broadcasters carry the factual baseline, and conservative outlets are read in full as primary evidence of what the right is being told, not as a garnish. The coverage gap section exists because that weighting has a predictable failure mode, and naming the failure mode is cheaper than pretending it does not exist.

Every figure, date, toll, price and officeholder in this review was checked against live web search: 34 load-bearing claims checked, four corrections, three mechanical and one judgment call. The mechanical corrections were a garbled partisan series on the Pew Islam question, restored from Pew's own release rather than a secondhand paraphrase; an officeholder check confirming that the attorney general referenced in the AP protest analysis was correctly described as the former officeholder, removed in April 2026; and a sourcing correction establishing that the Chinese satellite imagery story traces to the Wall Street Journal rather than to the outlet that aggregated it. The judgment call was the number of Iraqi border crossings closed, where the AP reported two and Al Jazeera reported three and named them. This review uses three.

Attribution discipline is one standard, applied without regard to which government is speaking. A government's claim about casualties, damage, attribution or fraud is reported as that government's claim, whether the government is American, Saudi, Iraqi or Iranian. Opinion columns and analysis pieces are labeled as argument and never presented as reported fact. Survey findings are attributed to the institutions that produced them. Where a procedural detail comes from a single outlet's reporting and was not independently re-sourced, that is said in the text.