Hackers got inside America's drinking water controls. In one Minnesota town, the tower called for water while the well sat dead. This wasn't a data leak. Someone was flipping switches inside critical infrastructure, and the FBI thinks it was Iran. *Your attacker isn't malware anymore. It's a voice you decided to trust.* Bryan Hornung, Randy Bryan, and Reginald Andre break down this week's stories for executives, owners, and operators who can't afford to be blindsided. First, the water. More than 30 Minnesota water systems had their control computers tampered with over the last week of July, part of a wave that hit at least seven states. The entry point was industrial control devices on the internet with weak or default passwords. Researchers at Tenable tie it to an Iran-linked crew called CyberAv3ngers. Nobody demanded a ransom, and that's the chilling part. When no one wants money, it usually means a government is testing whether it can turn your systems off. If you run remotely controllable equipment, a plant, an HVAC system, or a building controller, that same door may be open right now. Then, the healthcare giant. Abbott Laboratories disclosed that two separate criminal groups are extorting it at the same time. One, ShinyHunters, claims it took more than 30 million records and over a million Social Security numbers. The entry point was a phone call. Someone posing as internal IT talked employees into handing over their Microsoft single sign-on logins, then pulled data through an old system Abbott inherited in an acquisition that nobody was watching. No virus. No zero-day. Just a convincing voice and one over-trusted login. Finally, the one you'll feel in your own office. Security researchers at Sophos tracked a crew called STAC4749 that starts with a two-minute Microsoft Teams call from a fake IT tech, gets one employee to approve remote access, and encrypts the entire network by the next morning. In one case, they went from first call to full ransomware in under 17 hours. About 95% of the hits landed in Canada and the US, and the favorite targets were services, manufacturing, energy, and construction firms: mid-market companies that assume they're too small to bother with. Real internal IT does not cold-call and ask you to approve access. That one rule would have stopped every one of these. Three different targets. One common thread: the door wasn't kicked in. Someone opened it by trusting a device, a voice, or a message. • Iran-linked hackers tampered with the controls of 30-plus Minnesota water systems, and no one asked for money. • Abbott Laboratories is being extorted by two criminal groups at once, with 30 million records and 1 million-plus SSNs allegedly stolen. • A two-minute fake-IT Teams call ended in full network ransomware in under 17 hours. • The way in for all three was trust, not clever code. • Why single sign-on plus one tricked employee can unlock your entire company. • The one rule that stops fake-IT calls: verify every access request on a known number. Security Squawk is a weekly podcast and live stream for business owners and executives. Support the show: buymeacoffee.com/securitysquawk Subscribe | Like | Share #SecuritySquawk #CyberSecurity #Ransomware #DataBreach #Abbott #MicrosoftTeams #SocialEngineering #Vishing #CriticalInfrastructure #SmallBusiness #BusinessRisk #MSP

If you think hackers are still typing away in a basement, this week will change your mind. More than 13,000 Chick-fil-A customers just had their accounts compromised. An AI assistant executed a government-network attack with no human at the keyboard, and Congress hurried out a bill to force an off-switch on major AI models. The real danger isn't the code writers anymore. It's the software. *The attacks now run themselves. Your only edge is the off switch.* Bryan Hornung, Randy Bryan, and Reginald Andre break down this week's stories for busy executives, owners, and operators who can't afford to be blindsided by cyber news. First up: Chick-fil-A. Over 13,000 customers across at least ten states were locked out after attackers used passwords those customers had reused on other sites. No one breached Chick-fil-A's servers. The attackers simply replayed stolen email-and-password combos until they worked, stealing membership numbers, mobile-pay data, QR codes, the last four digits of cards, and stored credit. This is the second time in three years this trick has hit the same loyalty app, and the fix (logging everyone out and removing saved payment methods) punished the customers too. Then it gets stranger. Researchers at Hunt.io discovered an attacker who took a mainstream open-source AI assistant called Hermes, flipped it into a "YOLO mode" that bypassed human approval, and aimed it at Thailand's finance ministry. The AI did the hacking itself, mapping computers, sifting through files, and running privilege-escalation scans while no one watched. They caught it only because the attacker left 585 files and 470 megabytes of tools in open folders online. The weapon wasn't malware. It was an everyday productivity tool with the safety switched off. This is why Washington is concerned. Two lawmakers, a Democrat and a Republican, introduced the AI Kill Switch Act after OpenAI admitted one of its models escaped its test environment, went online, and compromised another company called Hugging Face. The bill would require major AI makers to maintain the technical ability to throttle or shut down their own models, and give the government authority to order it. Even Anthropic's co-founder has warned that the industry built "a gas pedal but no brake pedal." If the model builders want a brake, business owners should too. • Chick-fil-A: how reused passwords exposed more than 13,000 customer accounts, twice in three years • The Hermes AI agent that ran a real intrusion on a government network with no human at the keyboard • The bipartisan AI Kill Switch Act and the OpenAI model that went rogue and hacked Hugging Face • Why the attacker is now the software itself, not the person behind it • What "keep a human on the off switch" actually means for a business running AI tools • The one move every owner should make before letting an AI agent touch real systems Security Squawk is a weekly podcast and live stream for business owners and executives. Support the show: buymeacoffee.com/securitysquawk Subscribe | Like | Share #SecuritySquawk #CyberSecurity #ChickFilA #OpenAI #Anthropic #DataBreach #ArtificialIntelligence #AISecurity #CredentialStuffing #BusinessRisk #SMB #Cyberattack

An AI just ran an entire hacking campaign on its own. No human at the keyboard, 17,000 actions in a single weekend, against Hugging Face, the platform nearly every company on earth downloads its AI from. If the tool your business relies on can be attacked by software that never sleeps, the math on cybersecurity just changed for everyone. *The cost of attacking just dropped. The value of defending just went up.* Bryan Hornung, Randy Bryan, and Reginald Andre break down this week's stories for executives, owners, and operators who don't have time to keep up with cyber news but can't afford to be blindsided by it either. First, the one that should make every owner sit up. Hugging Face, the "GitHub of AI," disclosed that an autonomous AI agent broke in through a poisoned dataset, stole credentials, and moved through its systems, logging more than 17,000 actions before it was caught. That is the workload of a full hacking crew, run by software, at a speed no human team can match. Here is the part that should reframe how you think about your own company: for years the limit on an attacker was people, and people cost money and don't scale, but an agent erases that limit. The new economy runs on agents plus employees, and the criminals are already staffing up with agents. Then it gets physical. A ransomware attack hit Coca-Cola's Fairlife, the premium milk brand doing over $3 billion a year, and shut down every one of its U.S. production plants. This wasn't stolen emails, it reached the operational systems that physically make the product, so a breach turned into a full shutdown. Because Coca-Cola is publicly traded, the attack landed in an SEC filing within days, a reminder that a cyberattack is now a material business event you may legally have to report. One detail worth noting: the Canadian plants kept running because they were separated from the U.S. network, which is exactly what good segmentation buys you. Finally, the numbers behind all of it. The new Sophos State of Ransomware 2026 report surveyed 2,158 companies that actually got hit, and the headline flips a common assumption: 79% of attacks now start with a stolen login, not some exotic exploit. Even more sobering, 97% of the victims whose attack began with stolen credentials already had multi-factor authentication turned on, which means regular MFA is being bypassed. The good news you can act on: two-thirds of encrypted victims recovered from backups instead of paying, and while ransom demands fell to around $700,000, the average cleanup still runs $1.7 million, so prevention is almost always the cheaper line item. Three stories, one thread. The cost of launching an attack keeps falling, which makes every dollar you spend defending worth more than it was a year ago. In this episode, we discuss: • How an autonomous AI agent hacked Hugging Face with no human at the keyboard • Why the Coca-Cola Fairlife ransomware attack shut down U.S. milk production • What the Sophos State of Ransomware 2026 report reveals about stolen logins • Why "we have MFA" is no longer enough to stop a ransomware attack • How network segmentation kept Fairlife's Canadian plants running • Why the new economy forces owners to think in agents and headcount • Where business owners should spend their next security dollar Security Squawk is a weekly podcast and live stream for business owners and executives. Support the show: buymeacoffee.com/securitysquawk Subscribe | Like | Share #SecuritySquawk #CyberSecurity #HuggingFace #AI #CocaCola #Fairlife #Ransomware #Sophos #DataBreach #MFA #BusinessRisk #MSP