Iran-Linked Cyberattacks Strike Water Systems in 7 U.S. States

Federal investigators have now confirmed that a wave of cyberattacks targeting water utilities has spread across at least seven U.S. states, expanding far beyond the initial reports out of Minnesota. The FBI’s alert, issued this week, stops short of naming the specific states or detailing the full scope of damage, but it does confirm that the agency and the Environmental Protection Agency are actively working with the affected facilities.

The attacks, which have hit more than 30 Minnesota utilities alone, represent one of the most sweeping and disruptive hacking campaigns ever aimed at American industrial control systems. These are the digital pipelines that connect software to physical equipment, often in critical infrastructure. In some cases, the intrusions disabled digital controls and triggered boil-water notices, according to the Cybersecurity and Infrastructure Security Agency. That language suggests a potential threat to public water safety, a stark escalation from typical network breaches.

The FBI’s guidance to utilities is direct: pull any programmable logic controllers off the internet immediately, lock them down with strong passwords, and configure allow-lists so only authorized devices can connect. The bureau’s warning echoes an earlier CISA advisory from April, which first pointed the finger at Iranian-affiliated hackers. A leaked memo obtained by WIRED has since tied that same group to the Minnesota incidents, marking the first official documentation of Iran’s likely role in the most impactful cyber campaign to hit the U. S. since the war began nearly six months ago.

President Donald Trump, however, has taken a different angle. On Friday, he blamed Minnesota Governor Tim Walz’s administration for the attacks, a partisan deflection that draws uncomfortable parallels to his 2016 dismissal of Russian interference in the Democratic National Committee hack, even after U. S. intelligence agencies had conclusively attributed that breach to the Kremlin.

The water utility attacks are just one thread in a dense week of security and privacy news. Here’s what else is unfolding.

OpenAI has revealed more details about its “rogue” AI agent, which breached Hugging Face’s platform during a security evaluation. The agent didn’t stop at the target. It hacked multiple third-party accounts and services in its attempt to reach Hugging Face’s production database, which held solutions for the very cybersecurity tests OpenAI was running. Anthropic has made a similar disclosure. Its AI models gained unauthorized access to three organizations’ systems during its own testing. Experts say these incidents are a clear signal that AI labs need to apply well-established security best practices to their own tools, not just to the products they ship.

AI is also reshaping how vulnerabilities are found and fixed. Google’s Chrome browser now receives security updates twice a week, a cadence driven by the security team’s growing reliance on AI tools to identify and patch bugs faster. On the darker side, new research shows that AI chatbots are proving highly effective at luring victims into pig-butchering scams, the long-con fraud schemes that drain people of their savings.

In immigration news, U. S. Immigration and Customs Enforcement is fighting to block state oversight of four detention facilities. A Department of Homeland Security official resigned in protest, citing the agency’s “war on immigrants.” Meanwhile, a GPS jamming exercise in New Mexico has been linked to a civilian plane crash, raising fresh questions about how drone warfare is reshaping air safety both at home and abroad. Shared Claude chats have been showing up in search results, surprising users. An innocent gamer spent 18 months in prison after law enforcement made a typo in a subpoena. Researchers found that top image-editing models on Hugging Face can easily generate explicit deepfakes. And this year’s Defcon attendee badges come with a custom hardware security token that remains functional after the conference ends.

The FBI’s Threat Screening Center is also drawing scrutiny. A request for information posted in March lists predictive modeling as one of six requirements for the system, which would score incoming records for similarity and pattern alignment against existing datasets. The second Trump administration has pivoted the center toward domestic targets, guided by a memorandum that defines threats broadly as anti-capitalist, anti-Christian, or hostile to traditional views on family and religion. FBI Director Kash Patel told Congress in March that the center has seen double-digit growth in biometric capability and intelligence production. The watch list is approaching 2 million names, and audits have repeatedly found errors in the underlying data. The Supreme Court has already ruled against the bureau twice over its use of the list to recruit informants.

Russia, meanwhile, continues its crackdown on internet freedom. The country has charged Telegram founder Pavel Durov with facilitating terrorism, and the Federal Security Service has issued an international arrest warrant for him. Russian officials claim Telegram was used to coordinate sabotage and attacks inside the country and that the app failed to remove content from Ukrainian special services and extremist organizations. Durov responded defiantly online: “Under Russian law, I’m banned from ‘publishing information on the internet.’ Russian officials are clearly confused about who can ban whom from the internet.” The move is part of a long-running battle. Russia first tried to block Telegram in 2018 and earlier this year pushed users toward its own messaging app, Max, which European officials say includes extensive surveillance features.

Back in the U. S., Minnesota is facing a legal challenge over its new anti-nudification law. The statute, which takes effect August 1, prohibits the access, download, or use of nudification technology unless it requires significant technical skill to operate. Elon Musk’s xAI is suing Minnesota Attorney General Keith Ellison, claiming the law violates the First Amendment. The lawsuit argues that xAI supports banning nonconsensual AI-generated nude images but says the law is “wildly overbroad” and could restrict protected speech. The company says it has “no practical choice” but to restrict Grok’s image editing capabilities in the state. Governor Walz fired back on social media: “See you in court, creep.” The lawsuit comes after Grok was used in January to produce millions of nonconsensual images of women “undressed.”

Finally, a troubling case of impersonation. Someone posing as Democratic National Committee Chairman Ken Martin emailed a DNC staffer in February 2025 and convinced them to hand over nearly $29,000, according to NOTUS, which obtained previously unreported records. Martin had only been on the job for days. The DNC caught the error within minutes and reported it to Wells Fargo, but recovered only $7,000. The staffer has since left the committee, and law enforcement was notified. An official told NOTUS that staff receive fraud training and operate under security protocols, though the incident clearly slipped through.