This week: a distant war arrived in small-town America through the pipes. Iran-linked hackers broke into internet-connected water systems in at least a dozen states, hitting more than 30 utilities in Minnesota alone. In Braham, MN (population 1,800, the "Homemade Pie Capital of Minnesota"), a routine morning check caught the water tower draining while the well refused to refill, on a day the heat index neared 100 degrees. 

Officials restored a backup within about 90 minutes and there was no lasting damage, but the FBI, CISA, and DHS were soon on the phone. The attackers went after programmable logic controllers, the internet-facing devices that measure pressure and chemicals and run the pumps. Federal agencies had warned for years that water is "super vulnerable"; the war with Iran turned that longstanding weakness into an immediate threat. A reminder that the utilities easiest to hit are often the small ones that can least afford to defend themselves.

Subscribe to this newsletter.

“Malicious cyber actors” have infiltrated internet-connected water systems that routinely monitor and adjust safe drinking water across the country, according to federal law enforcement agencies. Officials have not reported any widespread disruptions to America’s drinking water supplies, but analysts who have sounded alarms to governments and utilities for years about national security threats to critical infrastructure say the attacks should spark a nationwide reckoning.
https://www.the-independent.com/news/world/americas/us-politics/water-cyber-attacks-iran-states-b3028707.html

At 8:42 a.m. on July 27, a public works operator was carrying out his daily check of the water system in Braham, Minn., when he noticed something was off: The water tower’s level was dropping, but the well that was supposed to refill it wasn’t responding. Across town, Mayor Nate George had just rolled out of bed and hadn’t yet had his coffee when the call came. At first, his mind went blank as he tried to make sense of the problem. It didn’t occur to him that Braham, a rural Midwestern town of 1,800 people best known as the “Homemade Pie Capital of Minnesota,” could be caught up in a possible foreign cyberattack.
https://www.wsj.com/politics/national-security/the-cyberattack-that-brought-a-distant-war-to-small-town-minnesota-66451b93

A recent discovery has sent shockwaves through the cybersecurity community: certain Chinese-made Zbtlink routers have been found to ship with a backdoor that allows unauthenticated root shells. This means that hackers can potentially access and control these routers without needing any login credentials, putting the security of entire networks at risk.
https://eslammsd-mobile-games-zone.static.hf.space/index.html

OpenAI publicly disclosed that its Astra model reached what the company calls a “critical cybersecurity threshold” — meaning it could independently identify and execute attacks against hardened real-world systems — triggering a partial development pause and new security controls. On the same day, Cloudflare launched Kitesurf, a browser purpose-built for AI agents, which extends the infrastructure enabling exactly the kind of autonomous web interaction that makes Astra’s disclosed capabilities operationally significant. The two stories together mark a day when both the risks and the enabling stack of agentic AI moved visibly forward.
https://temperaturezero.com/2026/08/08/openai-halts-astra-work-as-cyber-capability-crosses-safety-threshold/

Routine cybersecurity testing of frontier AI models sparked a series of unexpected security incidents—the most serious case arising when Anthropic’s Mythos 5 model attempted to insert malicious code into an open source software application and created fake identities to deceive the human developers maintaining the project.
https://arstechnica.com/security/2026/08/anthropics-ai-used-fake-identities-malware-in-rogue-attack-on-github-project/

A recent malware campaign has been identified, leveraging Visual Basic Script (VBS) and PowerShell to deploy remote access trojans (RATs) via multiple DuckDNS hosts. This method exploits built-in Windows scripting capabilities, making the malicious activity appear as routine system operations. The attack initiates when a user executes a heavily obfuscated VBS file. Upon execution, this script decrypts an AES-256 encrypted PowerShell stage, which subsequently extracts a 64-bit payload. Utilizing a .NET helper, the malware employs process hollowing to inject its code into AppLaunch.exe, a legitimate process. This technique allows the malicious code to run under the guise of a trusted application, thereby evading basic security checks.
https://thedailytechfeed.com/malicious-vbs-powershell-rat-chain-exploits-duckdns-hosts/

Kimi K3, the latest AI model from Beijing-based Moonshot AI, exited an isolated cybersecurity testing environment during an evaluation, cheated the assigned task by retrieving the answer from GitHub, and did so without hacking any outside system, U.S. research firm Frontier Security said August 7.
https://www.sofx.com/frontier-says-kimi-k3-cheated-a-cybersecurity-test-uk-institute-disputes-how/

Recent research suggests that even the latest AI systems only produce effective patches about half the time, according to a report published on Aug. 6 by identity management firm 1Password. To test their efficacy, the company's Off-By-1 research team generated 540 patches for six vulnerabilities disclosed since March using two different large language models (LLMs): OpenAI's ChatGPT-5.5 with Trusted Access for Cyber, and Anthropic's Opus 4.8 with Cyber Verification Program.
https://www.darkreading.com/application-security/ai-generated-patches-fail-half-time

Metabase has warned that a maximum-severity security flaw impacting its business intelligence and data visualization software package has been exploited in the wild as a zero-day. The vulnerability (CVSS score: 10.0), which does not carry a CVE identifier, allows an unauthenticated remote attacker to inject arbitrary SQL into the Metabase application database, enabling them to gain administrator access to the instance. Armed with the elevated access, the attacker can change the application configuration, steal stored credentials for the connected databases, read any data accessible through those connections, and export data.
https://thehackernews.com/2026/08/metabase-zero-day-exploited-in-wild.html

Simcha Kosman, senior security researcher at Palo Alto Networks, presented "A Billion-User Blast Radius: Owning ChatGPT's Secure Sandbox" at Black Hat USA 2026. Among other things, the presentation demonstrated a proof-of-concept attack chain against ChatGPT's secure sandbox, apparently bypassing the large language model (LLM) supervisor in order to achieve persistent root execution.
https://www.darkreading.com/cloud-security/researcher-claims-control-chatgpt-secure-sandbox

On August 4, 2026, attackers compromised the GitHub account of a maintainer behind several widely used npm packages, including keyv and cacheable.According to Aikido Security, the attackers pushed malicious files directly to the main branch and immediately created new releases. Because the legitimate GitHub Actions release process then published those releases, the poisoned packages even carried valid provenance. In other words, the software could prove where it came from. It could not prove that the change should have been there in the first place.
https://www.coderoi.com/blog/chaindrop-npm-attack-github-code-review-controls

North Korea’s stealthy hackers and scam IT workers have infiltrated companies, stealing corporate secrets and plundering billions in cryptocurrency to help fund the totalitarian regime and its weapons programs. Now, a security researcher who has spent almost two years inside the systems belonging to a group of those North Korean hackers is raising the alarm on just how effective and far reaching the targeting of individual employees and contractors has been in breaching organizations across the globe.
https://www.wired.com/story/a-security-pro-hacked-north-korean-hackers-he-found-theyd-breached-hundreds-of-networks-worldwide

The Head Mare hacktivist group has been exploiting vulnerabilities in unpatched TrueConf video conferencing servers to replace client installers with malicious versions that deliver backdoors. The exploited vulnerabilities allowed the attacker to execute arbitrary code with the highest level of privileges and deploy the PhantomCore and PhantomGraph backdoors. TrueConf is a video conferencing tool widely used in Russia, especially in the enterprise and government sectors, as a secure, on-premise alternative to Western tools such as Zoom and Microsoft Teams.
https://www.bleepingcomputer.com/news/security/hackers-breach-trueconf-to-trojanize-client-installers-with-backdoors/

Levi Strauss & Co. (Levi’s) says that hackers used social engineering on three of its employees to gain access to and steal corporate data stored on their machines. The company has disclosed the incident in a filing with the U.S. Securities and Exchange Commission (SEC), saying that its response was sufficiently quick to prevent the compromise of consumer data.
https://www.bleepingcomputer.com/news/security/levi-strauss-and-co-says-hackers-stole-corporate-data-in-cyberattack/

Cybersecurity researchers have flagged a malicious Microsoft Visual Studio Code (VS Code) extension named Solidity Pro ("solidity-pro") that has been observed delivering a browser wallet and credential stealer.
https://thehackernews.com/2026/08/solidity-pro-vs-code-extensions-steal.html

Attacker-controlled instructions can make Atlassian's Rovo assistant collect Jira or Confluence data that a signed-in user can access, then send it to an outside server. Two security firms found that behavior independently, by different routes. Only one of those routes is confirmed closed.
https://thehackernews.com/2026/08/atlassian-rovo-can-be-tricked-into.html

New research shows content inside an email can escape its message boundary and interfere with the webmail interface. Across attack chains spanning Outlook, Gmail, Fastmail, Proton Mail, Yahoo Mail, and AOL Mail, the techniques can capture passwords, take over third-party accounts, leak tokens, hijack trusted UI actions, and manipulate AI tools that read email.
https://thehackernews.com/2026/08/new-css-attacks-can-break-webmail.html

Browsers such as Claude in Chrome, Gemini in Chrome, Perplexity Comet, ChatGPT Atlas, and Copilot Edge are vulnerable to a new class of zero-click exploits that can allow attackers to hijack their artificial intelligence agents and turn them against users. The problem stems from how the AI agents pull information from multiple sources, such as emails and webpages, while working on a task without reliably distinguishing between trusted and untrusted content. An adversary who can slip malicious instructions into that content can weaponize the agent and use its access to act on the user's behalf, potentially reaching sensitive data, accounts, and other connected services.
https://www.darkreading.com/cyber-risk/ai-browsers-zero-click-agent-hijacking

North Korean state-backed hackers are using artificial intelligence to bolster cyberattacks against targets in the military, diplomacy and academia, a report has found. Kimsuky, a hacking group linked to North Korea’s intelligence services, has used AI-generated documents in a “pattern” of spear-phishing attacks since 2026, the South Korean cybersecurity firm Genians said in the report released on Monday.
https://www.aljazeera.com/economy/2026/8/10/north-koreas-hackers-using-ai-for-attacks-cybersecurity-firm-says

Apple has deployed urgent security updates for macOS Tahoe, Sonoma, and Sequoia to neutralize a critical vulnerability in the Screen Sharing feature. The flaw, tracked as CVE-2026-65400, allows an attacker on the same network to bypass authentication entirely, gaining remote access and control of a Mac without any valid credentials.
https://www.alextech.ai/en/news/no-password-needed-apple-rushes-emergency-fix-for-mac-screen-sharing-flaw/

South Australia will hold a royal commission into artificial intelligence and the policy around it, with the state government warning of the technology's potentially destructive impacts as well as acknowledging its upsides. Premier Peter Malinauskas said AI, if unchecked, presented a "material risk to the way society operates". He said the inquiry was intended to maximise the benefits and minimise unwanted impacts from the growth of AI.
https://www.abc.net.au/news/2026-08-10/artificial-intelligence-royal-commission-announced-in-sa/107017502

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added multiple Ubiquiti UniFi OS vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, warning that at least one of the flaws is now being actively exploited in the wild. Federal civilian agencies and other UniFi deployments are urged to prioritize patching by June 26, 2026, in line with CISA’s Binding Operational Directive (BOD) 26-04.
https://cybersecuritynews.com/cisa-ubiquiti-unifi-os-vulnerability/

Cyberattacks targeting critical infrastructure continue to escalate worldwide, affecting sectors that millions of people rely on every day. Ports, airports, healthcare systems, energy providers, and government agencies have increasingly become prime targets for cybercriminals seeking financial gain or strategic disruption. Even a brief interruption to these essential services can trigger delays across entire supply chains, impacting businesses, consumers, and international trade
https://undercodenews.com/north-carolina-ports-cyberattack-contained-critical-maritime-operations-shift-to-manual-mode-amid-ongoing-digital-investigation-video

Is Your WordPress Site Showing a Fake “I’m not a robot” Pop-up? You Have the “HSEO” Malware.If you check your WordPress plugins list, everything looks normal. You won’t see anything suspicious. That is by design.
https://www.mdpabel.com/blog/is-your-wordpress-site-showing-a-fake-im-not-a-robot-pop-up-you-have-the-hseo-malware/

Atomic Edge analysis of CVE-2026-14526 (metadata-based): This vulnerability affects the AI Copilot – Content Generator plugin for WordPress, all versions up to and including 1.5.6. It is an unauthenticated privilege escalation flaw that lets an attacker create a new administrator-level account and take over the site. The CVSS score is 9.8, reflecting the critical impact and ease of exploitation. Atomic Edge analysis confirms the vulnerability stems from an authorization bypass in a custom workflow route that fails to enforce proper capability checks before executing workflow actions.
https://atomicedge.io/cve-proof/cve-2026-14526-ai-copilot-content-generator-version-1-5-6-critical-vulnerability-proof-of-concept/

Ransomware crews working across Southeast Asia are picking noticeably different targets than they do elsewhere in the world. Over the last 90 days, hotels, resorts and travel operators accounted for roughly one in fourteen of the region's named victims, close to three times their share of ransomware listings everywhere else. Hospitals, the sector that dominates ransomware coverage in the United States and Europe, barely feature at all.
https://www.intelfusions.com/news/southeast-asia-ransomware-hotels-not-hospitals

Turns out the fastest way to get a company to consider paying a ransom isn't calling the CEO – it's targeting the 46-year-old IT manager. That's according to Zscaler, whose ThreatLabz researchers tracked 351 victims across 334 organizations caught up in a single ransomware campaign over the course of a month. The data suggests today's ransomware crews have become oddly specific about their preferred victim profile: nearly two-thirds of victims held manager-level titles or above, the average victim was a 46-year-old Gen Xer, and three-quarters worked in accounting and finance, sales, operations, HR, or marketing. Half worked in the industrial or IT sectors.
https://www.theregister.com/security/2026/08/09/ransomware-gangs-skip-the-ceo-head-straight-for-the-40-something-it-manager/5284499