Tariff Time For India & China's New AI Plan

Hi readers, happy Thursday! Today, we’re covering Trump’s India tariffs, the Minneapolis shooting, China’s AI push, (alleged) sleeper agents in Greenland, the sandwich man’s victory, Microsoft protests, and AI agent vulnerabilities.

“Just because things hadn't gone the way I had planned didn't necessarily mean they had gone wrong.” – Ann Patchett

Heavy Duties Hit Hyderabad

Donald Trump made good on his threat to punish India for buying up Russian oil yesterday, slapping the country with 50% tariffs. The tariffs will affect 70% of all Indian exports to the U.S. – key goods including pharmaceuticals, electronics, raw drug materials, and refined fuels will all go entirely un-taxed.

“All of us should follow the mantra of buying only ‘made in India’ goods,” said Indian Prime Minister Narendra Modi on Tuesday before the tariffs took effect. “Pressure on us may increase [from the tariffs], but we will bear it,” he added. Indian textile manufacturers have already been forced to close up shop due to a lack of sales, and exporters from other South Asian countries are already swooping in to sell their goods to U.S. clients.

Experts on U.S.-India relations say that the massive tariffs – among the steepest enacted by the Trump White House – will cause lasting damage on the two countries’ relationship, which the U.S. has carefully cultivated in order to counterbalance China’s influence in Asia. According to one anonymous Indian trade official, “Trump has blown it. The hard work between the two countries, which inherently did not trust each other but still managed to build a solid strategic relationship, is now at risk. It is going to take a long time to reboot, and it probably won’t happen until Trump is out.” In a TruthSocial post last month, Trump wrote, “I don’t care what India does with Russia. They can take their dead economies down together, for all I care.”

A Tragedy In The Twin Cities

Yesterday, a shooter opened fire at Annunciation Catholic Church in Minneapolis, killing two children and injuring 17 others (including 14 children). The children, students from the church’s Catholic school, were attending Mass at the church just before 8:30 a.m., at which time the shooter opened fire with a rifle, pistol, and shotgun before committing suicide. Fortunately, all of the injured are expected to survive, though four will need surgery.

Police are still investigating the shooting and have been unable to identify a motive in the case. The shooter had used a white marker to write various messages across their weapons before the attack. The writings paint the picture of an incoherent political agenda – they include antisemitic and racist language and also threaten violence against Trump. According to FBI Director Kash Patel, the shooting is being investigated as an “act of domestic terrorism and hate crime targeting Catholics.”

“Children are dead. There are families that have a deceased child … Don’t just say this is about thoughts and prayers right now. These kids were literally praying. It was the first week of school. They were in a church,” said Minneapolis mayor Jacob Frey. After it was revealed that the shooter might be transgender later in the day, Frey added: “Anybody who is using this as an opportunity to villainize our trans community has lost their sense of common humanity. We should not be operating from a place of hate for anyone. We should be operating from a place of love for our kids. This is about them.”

Stay Refreshed & Replenished All Day Long

Tripling Down On AI

• According to the Financial Times, China is planning to make a massive investment in its AI chipmaking industry next year. Citing anonymous sources, the Times wrote that one chip fabrication plant dedicated to making Huawei AI processors will likely be online by the end of 2025, and another two chip factories will be up and running by the end of 2026. Semiconductor Manufacturing International Corporation, the country’s leading chipmaker, is also planning to double its capacity to create its most advanced chips next year.

• While China’s AI industry is currently booming, it’s become bottlenecked due to a lack of domestic chipmaking capabilities – like most of the world, Chinese AI firms rely on chips from industry leader Nvidia, but the U.S. government has restricted China’s access to the firm’s most cutting-edge technologies. Now, the country is looking to increase its ability to create less-advanced but more efficient chips, which are being adopted by industry leader DeepSeek.

It’s A Sleeper Agent Summer

• It’s not a great week to be an American diplomat in Europe, apparently. After the U.S. Ambassador to France was summoned for reprimanding on Monday, the U.S.’s top diplomat in Denmark was also called in yesterday over allegations that at least three people connected to Trump were carrying out covert influence operations in the Danish territory of Greenland. In case you forgot, Trump has repeatedly floated the idea of the U.S. taking over Greenland, either by purchase or use of military force.

• According to Danish state broadcaster DR, one of those alleged Trump agents had collected a list of Trump supporters in Greenland with the goal of eventually recruiting them and building a secessionist movement within the territory. The other two have allegedly been nurturing relationships with local politicians, community leaders, and businessmen with the goal of weakening Greenland’s relationship with Denmark. The U.S. State Department responded to the claims by stating that “the U.S. government does not control or direct the actions of private citizens.”

Additional World News

• How Former Biden Officials Defend Their Gaza Policy (New Yorker)

• Kim Jong Un to join Putin and other leaders at China military parade (BBC)

• Japan denies giving away city to Tanzania after misinformation sparks anti-immigration backlash (CNN)

• Mexico suspends postal shipments to the US over latest tariff confusion (AP)

• Russian Missile and Drone Attack Kills at Least 8 in Kyiv (NYT, $) 

The Sandwich Man Is On A Roll

• Yesterday, federal prosecutors failed to secure a felony indictment against a man who threw a sandwich at a federal agent. Felony indictments require approval by a grand jury, but they’re usually easy to secure for two reasons: first, indictments require “probable cause” that the alleged crime occurred, and second, they’re only provided information on the case by prosecutors themselves. More remarkably, this is the second time in two days that federal prosecutors have failed to secure an indictment, with jurors rejecting felony assault charges in a case of a woman charged with assaulting an FBI agent.

• In the sandwich case, the defendant allegedly threw a hoagie at a Customs and Border Protection officer involved in Trump’s takeover of Washington, D.C. He allegedly also  called the officer and his colleagues “fascists” and told them, “I don’t want you in my city!” before being arrested. The fact that multiple juries made up of D.C. residents have rejected these indictments signals just how much the capital city doesn’t appreciate the Trump takeover, and more failures are likely on their way.

Microsoft’s Cloud Of War

• Earlier this month, a joint investigation by the Guardian and two Israel-based publications revealed that Unit 8200 (Israel’s military surveillance agency) has been using Microsoft’s Azure cloud platform to collect and store millions of phone calls made by Palestinians in Gaza and the West Bank. The data includes recordings of every call made by Palestinians living in those two territories, and is protected by custom security solutions created by Microsoft engineers. According to multiple sources from within Unit 8200, that data has been used to guide IDF airstrikes and military operations in Gaza.

• Over the course of this year, a growing number of Microsoft employees have been speaking out against their company’s cooperation with the Israeli military, demanding that it terminate all contracts with Israel’s government. That small protest movement reached a head today when seven people – including two current and three former Microsoft employees – brought their protest into Microsoft headquarters, staging a sit-in inside the offices of Microsoft president Brad Smith. All of the protestors were eventually removed from the premises, and Smith told the press that his company is “working every day to get to the bottom of what’s going on” with Israel’s use of its services.

Additional USA News

• CDC erupts in chaos after ousted chief Susan Monarez refuses to resign (Guardian)

• A Dark Money Group Is Secretly Funding High-Profile Democratic Influencers (Wired)

• John Bolton Inquiry Eyes Emails Obtained by Foreign Government (NYT, $)

• Colleges see significant drop in international students as fall semester begins (NPR) 

• The Trump administration promised a fourth wireless carrier — America got a hot mess instead (Verge)

• Nick Fuentes, America’s Next Top Racist (Atlantic)

New AI Fear Unlocked

• AI companies want you to believe that, someday soon, you’ll be able to ask an AI agent to grab you tickets to a basketball game, and those tickets will land in your inbox without you having to lift a finger. That does sound cool – and will probably become a reality within a few years – but the rise of AI agents comes with some serious security issues. 

• On Tuesday, AI startup Anthropic launched Claude for Chrome, an web extension-based AI agent which can carry out specific tasks (like grabbing dinner reservations or scheduling meetings) on behalf of users. In the leadup to that launch, the company conducted the usual security checks, and found that browser-using AI agents like Claude are particularly vulnerable to prompt-injection attacks. These attacks are when websites hide malicious instructions within their code, prompting the AI agent to do things like delete your entire email inbox or try to hijack your bank accounts.

• Over 123 cases, Anthropic found that prompt-injection attacks worked 23.6% of the time. The firm has since implemented a few security features reducing that rate to 11.2%, but in order to prevent the threat of catastrophic damage to users’ online information, it blocked Claude from accessing financial services, adult content, and pirated content. “I strongly expect that the entire concept of an agentic browser extension is fatally flawed and cannot be built safely,” said one AI researcher. Just last week security researchers showed that Perplexity's Comet, an agentic AI browser, could be used to steal users’ data through prompt-injection attacks hidden in Reddit posts. 

Additional Reads

• You lift bro? How America became a nation of exercise obsessives. (Vox) 

• Flesh-eating bacteria cases are rising. Climate change is to blame, say scientists (CNN)

• How to Survive Your Song Going Viral on TikTok (New Yorker)

• SpaceX Starship Finally Pulls Off a Successful Test Flight (Wired)

• Think you actually own all those movies you’ve been buying digitally? Think again (Guardian)

• How to maintain a healthy gut (Psyche)

Please consider supporting Daily Pnut, an independently operated and bootstrapped publication, and you'll receive our exclusive Sunday edition. Many thanks to everyone who already supports us!

Editor & Writer: Marcus Gee-Lim

Designer: Joe Stella