AI and IT News Recap: September 11, 2026: OpenAI Stops Selling Its 00 Pro Plan, Check Point Patches Two Critical VPN Flaws, and Miro Sells for 92% Less Than Its Peak

In short: Meta's Muse AI agent reached No. 2 on the US App Store with about 83,000 iOS downloads, a pace well behind Meta's previous app launches and behind ChatGPT's debut.

What it means for your business: Consumer agent apps are about to start showing up on employee phones, connected to personal accounts, doing tasks that may touch work data. That is a policy question more than a technology one, and it is worth a paragraph in your acceptable-use policy before it becomes a conversation after an incident.

My take: App Store rank is the most misleading metric in consumer software and it gets quoted the most. Eighty-three thousand downloads is a rounding error for a company Meta's size, and the Android number is the tell. The interesting competitive detail buried in the coverage is that the agent people actually seem excited about is Instinct, a much smaller product that works over text messages. If the winning form factor for consumer agents turns out to be SMS rather than an app, Meta's distribution advantage matters a lot less than everyone assumed.

Source: TechCrunch, "Meta's AI agent Muse is now the No. 2 app in the US"

🛡️ IT and security

Check Point patched two critical VPN flaws that need no password

Check Point released patches this week for two critical vulnerabilities in its gateway and firewall products that use VPN functionality. CVE-2026-85102 and CVE-2026-85103 both carry a CVSS score of 9.8, and both can be exploited without authentication to achieve remote code execution. The first is an improper validation of certificate data during VPN negotiation and affects Security Gateway and Check Point Spark Firewall when using site-to-site or remote access VPN. The second is a heap overflow in the VPN certificate ASN.1 decoding flow and affects the Security Management Server, Security Gateway and Spark Firewall.

Updates are out for versions R82.10, R82 and R81.20 across all affected products. Check Point found both flaws internally and says there is no evidence of exploitation in the wild, which is the good news and the reason to move now rather than later. If you cannot patch immediately, the vendor's mitigation is to disable implied rules for VPN and manually define access for UDP/500 and UDP/4500 to specific peer IP addresses, though that mitigation does not apply to locally managed Spark Firewall instances. Those should get the latest Jumbo hotfixes. Customers with Check Point LivePatch enabled receive the patches automatically. Context matters here: Check Point warned customers about two exploited zero-days in its VPN products earlier this summer, one of which was used in Qilin ransomware attacks.

In short: Check Point patched two unauthenticated remote code execution flaws rated 9.8 in its VPN-enabled gateway, firewall and management products, with no exploitation observed yet.

What it means for your business: If your perimeter is Check Point and you use VPN on it, this is a today task, not a next-maintenance-window task. Ask whoever manages your firewall for the firmware version in writing, and ask specifically whether LivePatch is enabled, because that determines whether you are already covered.

My take: The window between "vendor found it internally" and "someone reverse-engineers the patch" is the cheapest defensive time you will ever get, and almost nobody uses it. This is the second round of critical VPN problems at Check Point this year, and I want to be fair about that: every firewall vendor has had a run like this, and finding your own bugs is better than having them found for you. The pattern that should worry a business owner is not any one vendor, it is that the VPN concentrator has become the single most attacked box in a small network while also being the one nobody wants to reboot.

Source: SecurityWeek, "Check Point Patches Critical VPN Vulnerabilities"

IDScan confirmed the 153 million driver's license breach, in a notice search engines were told to ignore

Identity verification company IDScan has confirmed that an unauthorized third party accessed customer data in its cloud platform. The company learned of the incident on or around September 1 and published a breach notice on September 4, saying exposed information can include full names and driver's license or other government-issued ID numbers. It is offering free credit monitoring and identity protection. TechCrunch spotted the notice and reported that it had been published with a noindex directive, which instructs search engines not to list the page.

The backdrop is the part that makes this significant. Brian Krebs reported on September 1 that a dark-web platform called "Nexus" was advertising access to more than 153 million US and Canadian driver's license scans, along with 10 million ID cards, 3 million travel documents and 579,000 medical cards. Krebs verified samples with the consent of the people involved and traced the records back to IDScan. Lawsuits followed within days, while the company stayed silent publicly. IDScan's technology is used by car rental companies, retailers, financial institutions, cannabis dispensaries, gun shops and hospitality businesses to scan and authenticate government IDs. The Nexus service has since gone offline, though the criminals behind it very likely still hold the data.

In short: IDScan confirmed a breach of customer data in a September 4 notice published with a noindex tag, days after researchers tied it to a dark-web database of more than 153 million driver's license scans.

What it means for your business: If your business scans IDs at the counter, whether for age verification, rentals or check-in, you are handing a vendor the most sensitive identity documents your customers own. Find out today who that vendor is, how long they retain the images, and whether you can turn retention off. Most owners have never asked, and in many cases retention is a default setting rather than a requirement.

My take: The noindex directive is the detail I cannot get past. Publishing a breach notice while telling Google not to index it is technically compliance and practically concealment, and it tells you exactly how the company weighted its obligations. Beyond the bad look, this is a category of vendor most small businesses do not think of as a vendor at all. The ID scanner is a piece of equipment at the front desk, not a data processor in anyone's mental model, and that is precisely why the data sat somewhere nobody was watching.

Source: BleepingComputer, "IDScan confirms breach tied to 153 million stolen driver's licenses"

A misconfigured test server let attackers into Surfshark

VPN provider Surfshark disclosed a security incident this week. An internal test server became reachable from the internet after a misconfiguration, and a threat actor accessed it. The company discovered the exposure on August 31, initially treated it as low risk, then confirmed the scope on September 2 and moved to containment. The server held limited internal engineering material, described as parts of system binaries and internal configurations for certain services. Surfshark also found build-related credentials that had been committed into its code history and rotated them, saying those credentials did not provide access to user data or production systems.

The attacker also reached an isolated proxy server used for content accessibility optimization. Surfshark says no encryption keys, user identities, IP addresses or browser traffic were exposed, that no user data or VPN services were affected, and that no application or browser extension on a customer device was altered. The company has committed to an independent security audit of its broader infrastructure. Taken at face value, the segmentation worked: an engineering environment got breached and production did not.

In short: Surfshark disclosed that attackers accessed a misconfigured internal test server and a proxy server, and says no user data, encryption keys or production systems were affected.

What it means for your business: Two lessons, and neither is about VPNs. Credentials committed into source control are one of the most common ways small companies get breached, and they sit there for years. And a "test" environment that is reachable from the internet is a production environment whether you call it that or not.

My take: I want to give credit here, because the disclosure is unusually specific and the timeline is honest, including the admission that they first misjudged the severity. That is rarer than it should be. The substance is a reminder that the boring failure modes still dominate: a misconfiguration and secrets in git history, not some exotic attack. If you run any kind of development shop, go look at your repository history for credentials this afternoon. It is a free exercise and it almost always finds something.

Source: SecurityWeek, "Surfshark Systems Targeted by Hackers"

New Android malware encrypts files, steals your PIN, and then argues with you about the ransom

Researchers at Zimperium documented a new Android strain called Mantax Otax that bundles ransomware, spyware and outright harassment into one package. Indonesian operators distribute it through malicious APK files hosted outside Google Play, using phishing and social engineering to get people to install. Once on the device it requests Accessibility service permission, which grants it sweeping control, then pulls its command-and-control domain from GitHub and reports back the victim's location, carrier, Android version and device ID.

The encryption module only works against Android 9 and older, because the Scoped Storage protections introduced in Android 10 sharply limit what an app can reach. On those older devices it encrypts files with a victim-specific AES key, deletes the originals, appends a .enc extension, replaces local images with ransom notices, and opens a full-screen chat hosted on Firebase to negotiate payment. Zimperium found a misconfiguration in that Firebase server that exposed the attackers' conversations with victims. The spyware half works regardless of Android version: it can steal lock-screen PINs for persistence, read SMS messages and one-time passcodes, pull call logs, contacts, browsing history and Google account details, extract WhatsApp and Telegram messages through simulated Accessibility interactions, and abuse the MediaProjection API to take screenshots, record video and stream the screen in near real time.

In short: A new Android malware family called Mantax Otax combines file encryption on older devices with spyware that steals PINs, one-time passcodes and messaging app contents on any version.

What it means for your business: The one-time-passcode theft is the line that should get your attention, because it means SMS-based multi-factor authentication on a compromised phone protects nothing. If your team uses text-message codes for banking or email, this is the argument for moving to an authenticator app or a passkey. And the ransomware half only works on Android 9 or older, which is a good reason to find out whether anyone on your team is still carrying a phone that old.

My take: Sideloaded APKs remain the single biggest risk on Android and the one most business owners never think about, because it looks like a personal-phone problem right up until that phone reads your company email. The harassment feature is new to me and genuinely unpleasant, and it tells you the target is individuals rather than enterprises. The practical control is unglamorous: no sideloading on any device that touches work accounts, and get off SMS codes.

Source: BleepingComputer, "New Android malware encrypts files, steals data, and harasses victims"

A Conti ransomware operator got four years

A Ukrainian national has been sentenced to four years in prison for his role in Conti ransomware attacks. Oleksii Oleksiyovych Lytvynenko, 44, was arrested in Ireland in July 2023 at the request of the United States, extradited last year, and pleaded guilty to conspiracy to commit wire fraud in June 2026. He faced a maximum of 20 years. According to the Justice Department, he joined Conti in September 2021 as both an intruder and a developer, personally harmed at least 12 companies, controlled stolen data from eight US and four overseas victims, sent ransom notes, and coded a "loader," the malware component that delivers the tools an attack needs.

The scale of what he was part of dwarfs the sentence. The DOJ says Conti was used to attack computers and networks in 47 states, 31 foreign countries, the District of Columbia and Puerto Rico, and that the FBI estimated victim payouts exceeding