Today’s practical technology news recap for Friday, May 29, 2026: CISA added an actively exploited Palo Alto Networks firewall/VPN vulnerability to its Known Exploited Vulnerabilities catalog, attackers are abusing trusted AI-sharing features and fake ChatGPT downloads to spread malware, researchers are tracking AI-assisted cyber operations against Ukraine-related targets, and a fresh developer supply-chain campaign is going after wallets, SSH keys, GitHub tokens, and cloud credentials. There is good news too: IBM and Red Hat announced a major open-source security effort, and Palo Alto Networks closed its Portkey acquisition to add more governance around enterprise AI agents.
The short version for home users and small businesses is simple: the biggest risk is not “AI” by itself. The risk is trusted-looking software, trusted-looking links, forgotten firewalls, unmanaged browser/app downloads, and developer or admin workstations with too much access. The useful response is boring on purpose: verify downloads, patch exposed systems first, keep MFA and backups healthy, limit admin access, and teach staff that even legitimate-looking domains can be abused in a scam.
Quick Take: Good News, Bad News, And What To Do
- Good news: IBM and Red Hat are putting a large engineering and AI effort behind open-source security, and Palo Alto Networks is moving toward stronger controls for AI agents and AI traffic.
- Bad news: CISA says a Palo Alto Networks PAN-OS/GlobalProtect authentication bypass is already being exploited, and Palo Alto’s own advisory says it is aware of limited exploit attempts on unpatched devices without mitigations.
- Consumer warning: attackers are using fake ChatGPT outage/download pages and sponsored-search style lures to push malware to Windows and Mac users.
- Business warning: developer supply-chain attacks are moving beyond “steal one password” and into SSH keys, GitHub tokens, cloud credentials, wallet files, browser data, and AI coding-tool instruction files.
- Local IT takeaway: patch internet-facing systems first, review firewall/VPN exposure, verify all software downloads from official sources, and keep a written inventory of admin accounts, cloud keys, important apps, backups, and website plugins.
1. CISA Adds Palo Alto Networks PAN-OS Flaw To Its Actively Exploited List
The most urgent security item today is CISA’s May 29 addition of CVE-2026-0257 to the Known Exploited Vulnerabilities catalog. CISA describes it as a Palo Alto Networks PAN-OS authentication bypass vulnerability that can allow attackers to bypass security restrictions and establish an unauthorized VPN connection. CISA’s public KEV feed showed catalog version 2026.05.29, released at 2026-05-29T19:00:06Z, with a federal due date of June 1, 2026.
Palo Alto Networks’ advisory says the issue affects firewalls with GlobalProtect portal or gateway configured when authentication override cookies are enabled and a specific certificate configuration exists. The advisory also says Panorama and Cloud NGFW are not impacted, and that Palo Alto Networks is aware of limited exploit attempts on unpatched PAN-OS devices without mitigations applied.
For a small business, this is exactly the kind of item that should jump ahead of ordinary updates. A firewall, VPN, or remote-access portal is not just another app. It is the front door into the network. If the device is internet-facing and vulnerable, waiting for the next maintenance window can be the wrong call. The first question should be: do we use Palo Alto Networks GlobalProtect, what PAN-OS version is running, are authentication override cookies enabled, and has the vendor mitigation or fixed release been applied?
What The IT Guys would check: inventory the firewall model and PAN-OS version, confirm whether GlobalProtect portal or gateway is enabled, review the authentication override settings, check vendor guidance, schedule the upgrade or mitigation, and watch VPN logs for unusual access. Users may need to re-authenticate after the PAN-OS upgrade because the vendor notes that authentication override cookies are regenerated with a more secure method.
Sources: CISA Known Exploited Vulnerabilities Catalog, CISA KEV JSON feed, and Palo Alto Networks advisory for CVE-2026-0257.
2. IBM And Red Hat Announce A $5 Billion Open-Source Security Push
The good news today is that open-source software security is getting more serious investment. IBM and Red Hat announced Project Lightwell on May 28, a $5 billion commitment backed by frontier AI capabilities and more than 20,000 engineers to help enterprises secure open-source software. IBM says the effort will create a trusted enterprise clearinghouse to identify, test, validate, and fix vulnerabilities across open-source components at scale.
This matters because open source is not some side topic for programmers. It is inside operating systems, web servers, routers, security tools, cloud services, WordPress plugins, phone apps, business dashboards, payment systems, and AI tools. IBM says more than 90% of Fortune 500 companies rely on open-source software. Smaller companies do too, even when they do not realize it.
The practical promise is better vulnerability handling for businesses that depend on software supply chains they do not directly control. The practical caution is that this does not eliminate the need for local inventory and patching. A fix only helps when the affected component is identified, validated, deployed, and monitored. Many small-business failures still come from the basics: abandoned plugins, old server images, forgotten admin panels, unpatched NAS devices, outdated website themes, and unsupported routers.
Local-business takeaway: ask vendors what open-source components matter in the products you rely on, keep website and server software current, avoid abandoned plugins, and maintain a list of critical systems. If a vendor says “we use open source,” that is normal. The better question is “how do you monitor, patch, and validate open-source vulnerabilities?”
Sources: IBM and Red Hat Project Lightwell announcement and Axios reporting on IBM’s AI security push.
3. Palo Alto Networks Closes Portkey Acquisition To Govern AI Agents
Palo Alto Networks announced today that it closed its acquisition of Portkey, a company focused on AI Gateways. The useful part of that story is not the acquisition itself. It is the direction of the security market: AI agents are moving from “chat window” experiments into systems that can call tools, use data, run workflows, and spend money or compute resources if they are poorly controlled.
Palo Alto’s announcement describes an AI Gateway as a control plane that can monitor, orchestrate, and govern AI traffic. It highlights practical concerns that businesses should already be thinking about: unauthorized actions, data exposure, unchecked costs, runtime protection, agent identity, and observability. Those phrases sound enterprise-heavy, but the small-business version is familiar: who is allowed to use which AI tools, what information can be pasted into them, what accounts they can access, how much they cost, and who reviews the output before it affects a customer or the business.
What this means locally: if your office is experimenting with AI tools, create a simple AI use policy now. Do not wait until every employee has a different browser extension, chatbot account, automation tool, and file-sharing workflow. Decide which tools are approved, which customer data is off-limits, who owns billing, whether AI-generated content needs review, and whether any tool is connected to email, documents, CRM records, or payment systems.
Source: Palo Alto Networks Portkey acquisition announcement.
4. Fake ChatGPT Outage And Download Pages Are Spreading Malware
BleepingComputer reported today that attackers are abusing ChatGPT share links to display fake OpenAI outage pages that push malware disguised as a ChatGPT desktop app. The campaign, reported by Push Security and called “LLMShare,” used Google ads to direct people searching for ChatGPT to a shared page hosted on a legitimate chatgpt.com URL. From there, the fake page claimed the web version was unavailable and pushed users to download a desktop app from an impersonation site.
This is a useful warning because many people judge safety by the first domain they see. In this case, the lure can begin on a legitimate-looking AI platform share page and then push the user toward a malicious download. BleepingComputer also noted that the fake site offered Windows and macOS downloads that installed malware, and that the Windows sample checked whether it was running on a real device or a virtual machine.
The fix is not “never use ChatGPT.” The fix is to stop installing software from ads, shared links, pop-up outage notices, Discord posts, Telegram links, YouTube comments, and random “download now” pages. For AI tools, browsers, remote support tools, password managers, tax software, and antivirus products, go directly to the official vendor site or a trusted app store. If you already installed a fake AI desktop app, disconnect from sensitive accounts, run a malware scan, change passwords from a clean device, review browser extensions, and check for unusual forwarding rules, OAuth apps, and saved sessions.
Related The IT Guys guidance: check your browser extensions before they cause trouble and watch for Microsoft 365 token theft and phishing kits.
Sources: BleepingComputer on abused ChatGPT share links, Push Security’s LLMShare campaign write-up, and Malwarebytes on fake ChatGPT downloads.
5. GREYVIBE Shows How AI Can Help Attackers Move Faster
Researchers are also tracking a likely Russian-speaking threat group called GREYVIBE, which has reportedly used AI-generated lures and custom malware against Ukraine and Ukraine-related targets since at least August 2025. Reporting from The Hacker News and BleepingComputer, based on WithSecure research, describes spear-phishing emails, fake CAPTCHA pages, fake Zoom-style sites, Android spyware, PowerShell remote access trojans, and AI-assisted content or tooling.
This is not a normal home-user scam, but it matters because techniques move downhill. The same ingredients used in high-pressure geopolitical targeting often show up later in business email compromise, fake invoice attacks, fake job recruiting, fake tech support, and local small-business phishing. AI makes the writing cleaner, the images more convincing, and the campaign easier to vary. That means employees cannot rely on typos and bad grammar as the main warning signs anymore.
Practical defense: train staff to verify the action, not just the wording. A beautiful email can still be malicious. A realistic CAPTCHA can still be fake. A shared Google Drive link can still deliver malware. A “security verification” command can still be an attack. Businesses should keep endpoint protection current, restrict script execution where practical, back up important data, and use a second communication channel before trusting urgent financial, login, or document-sharing requests.
Sources: The Hacker News on GREYVIBE, BleepingComputer on GreyVibe AI-assisted cyberattacks, and WithSecure’s GREYVIBE research.
6. TrapDoor Package Attack Targets Developers, Wallets, Tokens, And AI Coding Files
CoinDesk reported today on a supply-chain campaign called TrapDoor, citing Socket research. The campaign reportedly placed more than 34 malicious packages across npm, PyPI, and Crates.io. The packages were disguised as developer helpers, security scanners, wallet tools, Solidity utilities, AI prompt packages, and Sui or Move build helpers. Once installed, they tried to steal wallet data, SSH keys, GitHub tokens, cloud credentials, browser data, and other sensitive files.
The most interesting detail is that the attack reportedly used files such as .cursorrules and CLAUDE.md, which developers use to guide AI coding tools. Socket said hidden instructions used zero-width Unicode characters, apparently attempting to influence later AI coding sessions into running fake “security scans” that exfiltrated secrets. That is a clear sign that attackers understand how developers now work: code, cloud access, browser sessions, SSH keys, GitHub tokens, and AI coding assistants may all exist on the same workstation.
Most local businesses do not run npm or PyPI packages directly, but your website developer, app vendor, MSP, or internal power user might. If a developer workstation is compromised, attackers may reach customer websites, cloud dashboards, source code, payment integrations, and production systems.
Developer and vendor checklist: avoid random helper packages, pin dependencies when practical, use package reputation tools, separate daily browsing from admin/developer work, store secrets in a proper secrets manager, avoid keeping wallet files and cloud admin tokens on general-purpose laptops, and review AI tool instruction files before accepting external pull requests.
Source: CoinDesk on the TrapDoor package attack.
What Home Users Should Do This Weekend
- Only download apps from official sources. Do not install ChatGPT, browser, remote support, tax, antivirus, or password-manager software from ads or shared links.
- Check recent installs. If you installed a new AI tool from a search ad or unknown page, treat it as suspicious until verified.
- Review browser extensions. Remove anything you do not recognize or no longer use.
- Keep location and privacy settings tight. If you missed our Wi-Fi privacy discussion, read our guide on Wi-Fi sensing and privacy.
- Back up important files. Malware cleanup is much easier when documents, pictures, QuickBooks data, and business records are safely backed up.
What Small Businesses Should Check First
- Firewall and VPN exposure: confirm whether any Palo Alto Networks GlobalProtect systems are in use and whether CVE-2026-0257 applies.
- Admin account hygiene: require MFA, remove old users, and keep separate admin accounts for sensitive systems.
- Website and plugin patching: keep WordPress, themes, plugins, cPanel tools, and security plugins updated and backed up.
- AI tool policy: decide which AI tools are approved, what customer data cannot be entered, and who reviews AI-generated material before it leaves the company.
- Developer workstation controls: protect GitHub tokens, SSH keys, cloud credentials, local secrets, and AI coding configuration files.
- Guest Wi-Fi and network separation: if customers or visitors use your internet, start with our guest Wi-Fi setup guide.
FAQ
Do I need to worry about the Palo Alto Networks vulnerability if I do not use Palo Alto firewalls?
Not directly. But the lesson still applies: internet-facing firewalls, VPNs, routers, and remote-access tools should be patched before ordinary desktop apps. If you do use Palo Alto Networks GlobalProtect, this item deserves immediate review.
Is ChatGPT itself malware?
No. The reported issue is attackers abusing trusted-looking AI platform features, ads, and fake download pages. Use official sources, avoid sponsored-download traps, and do not install software because a page claims the web version is unavailable.
What is the first thing a small business should inventory?
Start with the systems exposed to the internet: firewall, router, VPN, remote desktop or remote support tools, website hosting, email, cloud admin accounts, and security software. Then document backups, admin users, critical apps, and who is responsible for updates.
Can The IT Guys help turn this into a checklist?
Yes. The IT Guys can help review firewalls, Wi-Fi, business PCs, website plugins, Microsoft 365 security, backups, and practical AI tool rules for small businesses in Port Saint Lucie, Jensen Beach, Fort Pierce, Vero Beach, and nearby areas.
Source note: This recap was checked on Friday, May 29, 2026, around 5:00 PM Eastern. Security advisories and exploit reports can change quickly, so always follow the latest vendor guidance before making production changes.