Steve Miller's Blog

A digital workshop for systems-minded tech commentary.

Author: steve

  • Ransomware Victims Report Threats Against Employees’ Families in a New Survey: Your Incident Response Plan Needs a Personal Safety Chapter

    Ransomware Victims Report Threats Against Employees’ Families in a New Survey: Your Incident Response Plan Needs a Personal Safety Chapter

    Search for “ransomware physical threats employees families” this week and you will find a burst of coverage built mostly on one source: Semperis’s 2026 Ransomware Risk Report. The numbers are alarming. They are also self-reported survey figures from a security vendor, and the reporters who looked at them said they could not independently confirm them. I think the topic deserves a place in your incident response planning anyway, and this post explains both halves of that.

    Ransomware Physical Threats to Employees’ Families: What the Survey Found

    First, who produced the data. Semperis, which sells identity recovery and crisis response services, says in its report that it partnered with research firm Censuswide, which surveyed “1,200 IT and security professionals” in nine countries and several industry sectors in the first half of 2026. The report says it “surveyed only organizations that were successfully attacked.” That means the percentages describe organizations that were already ransomware victims, not companies in general.

    Here are the figures as the report states them:

    • “Threatening texts, letters, or packages”: 57%.
    • “Threats against family or friends”: 37%.
    • “Physical intrusion”: 19%. “Abduction”: 17%. “Bodily harm”: 13%. “Stalking”: 6%.
    • On the effects: “doxxing,” 25%, and “swatting,” 23%. Swatting is defined in the report as “harassed by fabricated emergency reports that resulted in law enforcement being dispatched to their location.”
    • “More than half (53%) of respondents reporting that ransomware attackers carried through on threats in the past year.”

    The report also says “41% of respondents said they had been the target of physical or psychological intimidation.” Semperis’s October 14 press release gives slightly different versions: “40% of attacks involved psychological or physical threats against staff” and “24% of victims or family members were swatted.” The report’s body says 23% for swatting. I cannot tell which is a typo or a rounding choice, so treat these as roughly 40% and roughly a quarter. The press release also says “nearly 25%” faced psychological intimidation, while the report’s list of attacker threats shows psychological intimidation at 23% and physical threats at 17%. The denominators are not always clear. For example, the 57% figure appears to be a share of people who were threatened at all, as Dark Reading describes it, not a share of all respondents.

    CSO Online’s coverage repeats the survey figures, including that “37% said employees’ friends or family members had been threatened,” and notes the survey’s sample and scope. It does not add independent evidence of its own.

    What the Data Cannot Tell Us

    Dark Reading is the most cautious source, and its caution matters. It writes that Semperis’s study “relied on self-reporting, and it couldn’t corroborate these alarming figures, nor could Dark Reading.” It adds that “Dark Reading failed to find even a single documented case of violence perpetrated in service of a ransomware attack.” When it asked Semperis for a case study, Semperis pointed to a Europol release on so-called violence-as-a-service, which “did not connect violence-as-a-service with ransomware.” A Semperis incident response director told Dark Reading that physical harm “does happen” in “some way, shape, or form.”

    Dark Reading also notes year-over-year confusion: it says 40% of participants in Semperis’s 2025 report reported physical threats, while only 17% of 2026 respondents said they had been subject to physical threats. It also reports related trends in other crimes, such as violent “wrench attacks” in cryptocurrency theft, and says that “What’s true of cryptocurrency is not necessarily true of enterprise cyberattacks.” I am not citing those numbers here beyond noting that Dark Reading frames them as adjacent, not proof.

    So the responsible reading is this. One vendor survey says a large share of attacked organizations report personal threats, including against families. Independent verification of violence is missing, and prevalence in the wider population of companies is unknown. I would not tell your executives that gangs are now routinely stalking employees’ families. I would tell them that credible-enough reports of family-directed threats exist, and that a plan for them costs little.

    What the Report and Coverage Recommend

    Semperis’s report advises documenting “roles not just within the cybersecurity team, but across the entire enterprise,” establishing out-of-band communications, and including “procedures for how to respond to incidents that involve places and people outside the organization, such as threats against family members,” including reaching out to local and federal law enforcement in advance. CSO Online’s article says that employees should have “a confidential channel to report intimidation” and that HR, legal, cybersecurity and corporate security teams should know when to escalate to law enforcement. It also quotes the CEO of the consultancy Primus Partners saying the organization should “minimize information retention and access to sensitive personal information of employees.” Those recommendations come from vendor, media and consultant voices, not from regulators.

    My Recommendation: A Personal Safety Chapter

    This section is my opinion. The checklist is my own, built on the sources above, and no source requires it.

    Picture this hypothetical. During an incident, a member of your on-call team tells their manager that someone has contacted a relative. If the runbook has nothing to say, the manager improvises, and the employee may feel alone with the problem. A short chapter prevents that.

    1. Name an owner. Pick one person from HR, legal or corporate security who handles personal-safety reports during incidents.
    2. Give staff a private reporting path. Make it clear how to report any contact or threat, with no penalty and no need to prove it first.
    3. Pre-build the law enforcement contacts. Know whom you would call locally and federally before you need them.
    4. Decide support up front. Agree in advance what help you can offer affected staff, such as counseling or security advice, and who can approve it.
    5. Reduce exposed employee data. Review what personal information about staff sits in company systems and public profiles, and cut what you do not need.
    6. Separate safety from the ransom decision. Document who decides how to respond to threats against people, and keep that apart from the technical recovery plan.
    7. Rehearse it. Add one personal-threat inject to your next tabletop.

    The Bottom Line

    The evidence is one vendor survey with self-reported answers and independent coverage that could not verify it. The planning cost is small, and the downside of having no plan is borne by people. That is why I would add the chapter, while being plain about what we do and do not know.

  • Drone Strikes Disrupted Yandex Services After Three Data Center Hits in a Week: Why Physical Attacks Belong in Your Disaster Recovery Threat Model

    Drone Strikes Disrupted Yandex Services After Three Data Center Hits in a Week: Why Physical Attacks Belong in Your Disaster Recovery Threat Model

    Most disaster recovery plans picture a fiber cut, a flood or a power failure. This month’s reporting on Yandex adds a scenario that many plans do not include: physical attack on a data center. I want to be careful here. This is a war, the facts come from companies and governments with a stake in the story, and I will attribute each claim to the source that made it. The reason I am writing about it is narrower: what the reports describe about data center physical attack disaster recovery is worth a look from anyone who runs, or buys, hosted infrastructure.

    What Yandex and the Press Report

    Yandex, the Russian technology company, described the first two incidents in an October 9 statement: “On the night of October 7th to 8th, the Yandex data center in Sasovo was seriously damaged in a drone attack. It has been completely shut down. This morning, another of our data centers, in Kaluga, was attacked. Part of its infrastructure is out of service.” It also said “No employees were hurt in either Sasovo or Kaluga” and that “it is very difficult to say when the equipment at the data centers will be back up and running.”

    The statement does not say who launched the drones. The attribution comes from elsewhere. The BBC reports that Ukraine’s defence ministry “confirmed the data centre attacks,” saying they could “force Russia to divert resources to repair the damaged capacity,” and that President Volodymyr Zelensky said the attacks were in response to weeks of Russian strikes on Ukrainian data centers. Zelensky is quoted by The Record as telling reporters: “We always respond in kind. You all know that they have attacked and continue to attack our data centers. We responded.” Those are Ukrainian statements, reported by the BBC and The Record. I cannot independently verify them.

    The BBC adds that “Moscow has justified the attacks by saying the data centres were used by Ukrainian military intelligence,” while Ukraine “accuses it of deliberately targeting civilian infrastructure.” That describes the Russian and Ukrainian positions on strikes against Ukrainian data centers. I am not taking a side on either claim.

    The third incident came days later. The BBC reported that Yandex “confirmed on Telegram that its data centre in Vladimir had been damaged and that operations had been ‘completely suspended’,” with some services unavailable to users. The BBC published that on Sunday, October 11, calling it the third attack that week.

    What Was Affected

    The Record from Recorded Future News reports the disruption reached beyond Yandex’s own apps. It says the attacks “disrupted Yandex Cloud, which provides computing infrastructure to thousands of businesses, as well as some of the company’s own services and third-party platforms that depend on its servers.” Citing the Russian internet watchdog, it reports users “have experienced problems accessing email, booking taxis, ordering food and making payments through Yandex Pay.” It adds that “The full extent of the disruptions remains unclear” and that Yandex has not said how many customers were affected.

    Two reporting details differ, so I will flag them. The BBC says two of Yandex’s five data centers were struck in its first report, while The Record says three of six had been targeted by Sunday. I cannot reconcile the totals from these sources, so I would not repeat either as a company-wide fraction. Also, The Record says the Sasovo site “reportedly” houses two of Yandex’s three AI-training supercomputers, and that Yandex has not disclosed whether any were damaged. The BBC likewise says it is “unclear whether the supercomputers survived.”

    Some of the most relevant details for IT teams come from The Record. It reports that after the latest strike, “the company advised cloud customers to consider alternative recovery plans rather than relying solely on switching between parts of Yandex Cloud’s own infrastructure.” It says Russian state media reported agreements with competing providers Selectel, K2Cloud and VK Cloud to help customers move data and services. And it reports that Yandex updated its terms of service “to exclude liability for failing to meet contractual obligations because of drone strikes and other attacks,” which could limit customers’ ability to seek compensation. I did not independently confirm those terms.

    Data Center Physical Attack Disaster Recovery: My Takeaways

    This section is my opinion and recommendation. The sources do not say any of it, and I am not drawing conclusions about the war.

    The reporting does not tell us how Yandex designed its sites, so I will not guess about its architecture. What I can say is what the advice to customers implies. If your recovery plan assumes you can fail over inside one provider’s infrastructure, a regional or multi-site event may test that assumption. Most of us will never face drones, but the same planning question applies to any event that takes several nearby facilities offline at once.

    Here is what I would review:

    1. Ask where your workloads and backups physically are. “Two sites” can still be a single geography decision. Ask for locations by region, not just availability zone names.
    2. Test a failover that leaves your primary provider. If you only practice switching inside one cloud, you have not tested the case where that provider is degraded across several sites.
    3. Keep a restore path you control. Hold backups somewhere your provider’s incident cannot reach, and practice restoring from them.
    4. Read the liability and SLA terms. The Record’s report that a provider changed its terms to exclude attack-related failures is a reminder to know what you could claim.
    5. Map hidden dependencies. The Record notes that third-party services hosted on Yandex infrastructure were also affected. List which of your vendors run on the same provider as you.
    6. Set your own decision triggers. Decide in advance what level of provider degradation triggers a move to your alternate plan, and who can approve it.

    Picture this hypothetical. Your main application runs in two zones of one provider’s region, and your backups sit in the same provider. A regional event degrades both zones. The plan says “fail over to zone B,” but zone B is part of the problem. That is the kind of gap a single-provider test will not reveal.

    The Bottom Line

    The facts that hold up across these sources are limited: three Yandex data centers were reported attacked by drones between October 7 and 11, two were reported shut down, services were disrupted, and Ukrainian officials have publicly described the strikes as retaliation. Everything beyond that is either unclear or opinion. For infrastructure teams, my takeaway is modest and practical: test the recovery path that does not depend on a single provider’s geography.

  • When the Zero Trust Proxy Goes Down, Some Microsoft 365 Users Go Down With It: Planning for Your SSE Provider’s Bad Day

    When the Zero Trust Proxy Goes Down, Some Microsoft 365 Users Go Down With It: Planning for Your SSE Provider’s Bad Day

    On October 13, some Microsoft 365 users could not reach their mail, files and meetings, and the cause was not Outlook or Teams themselves. Microsoft’s own incident notes point to its Global Secure Access service, the cloud network layer many organizations use to route traffic to Microsoft 365. Search for “Microsoft Global Secure Access outage Microsoft 365” and you will find coverage of exactly this event. That shorthand is useful as long as you keep the narrow scope in mind. This post lays out what Microsoft’s notices actually say, then what I would change in a plan that depends on a security service edge (SSE) provider.

    What Microsoft’s Notices Say

    Microsoft tracked the event as incident MO1505952. The notice text, as republished by 365Fanatic from a West Europe tenant and in the University of Pennsylvania’s mirror of Microsoft’s incident reports, reads like this:

    • Title: “Users may be unable to access multiple Microsoft 365 services using Global Secure Access (GSA).”
    • Services listed: “Impacted services may have included, but not been limited to: Outlook, Outlook on the web, OneDrive, SharePoint Online, Microsoft Teams, Windows 365 Cloud PCs.”
    • Root cause: “A section of GSA infrastructure was performing below acceptable performance thresholds, affecting connectivity to Microsoft 365 services and resulting in impact.”
    • Start and end: “Tuesday, October 13, 2026, at 7:40 AM UTC” to “Tuesday, October 13, 2026, at 6:31 PM UTC.” That is 3:40 AM to 2:31 PM ET, or about 10 hours 51 minutes.

    The cause Microsoft names is in Global Secure Access itself. Nothing in these notices attributes the problem to Microsoft Entra ID or Exchange Online, and I am not claiming either way beyond what Microsoft says. I am also not going further on root cause. Microsoft has published a post-incident report, according to the status entries, but I could not read its contents.

    Microsoft Global Secure Access Outage, Microsoft 365 Impact: Narrower Than the Headlines

    This matters for anyone deciding how worried to be. The impact was scoped to users “using GSA,” not to everyone on Microsoft 365. IncidentHub, which recorded Microsoft’s updates, writes that its updates “do not say Microsoft 365 was unavailable for users who were not using GSA.”

    The geography changed during the incident. Early updates said “Some users in Germany and the United Kingdom attempting to access Microsoft 365 services listed in the More info section using GSA may be impacted.” Microsoft’s final notice narrows that to “Some users in Germany.” IncidentHub’s timeline says the scope was narrowed to Germany only in the 17:26 UTC update. So “UK and Germany” is accurate for most of the incident, and “Germany” is what Microsoft’s final notice states.

    The start time also moved. IncidentHub notes that updates before 17:26 UTC stated an 08:00 UTC start, which Microsoft later revised to 07:40 UTC. Aggregators therefore disagree on duration. I use Microsoft’s final start and end times.

    One more useful detail. Microsoft’s updates show it first tried “performance and traffic optimizations,” then “redirecting traffic from the underperforming infrastructure,” then “targeted restarts.” A status entry also says “As a workaround, users may disable GSA to access the affected Microsoft 365 services until the resolution of this incident.” IncidentHub says that workaround first appeared at 18:20 UTC, about 11 minutes before the stated end. Other trackers time that entry slightly earlier, so treat the exact minute as approximate. Either way, it appeared late in a window of almost 11 hours.

    Planning for Your SSE Provider’s Bad Day

    Everything from here down is my own opinion and recommendation. Microsoft and the sources above do not say any of it.

    Here is my view of the lesson. Moving to zero trust and a cloud security edge removes some old single points of failure, such as one VPN concentrator. It can also add a new dependency. If your users reach Microsoft 365 through a cloud service edge, then that service’s health is part of your Microsoft 365 availability. This incident does not show that every organization has that exposure. It shows that organizations that do route through GSA were exposed to it.

    Picture this hypothetical. Your help desk gets a wave of “can’t open Outlook or Teams” tickets from one region. Your tenant dashboard shows no Exchange or Teams problem, and half the team spends the first hour troubleshooting endpoints. Nobody thinks to check whether the security edge is the common factor.

    What I would put in place:

    1. Know which traffic depends on the edge. Document which user groups, regions and Microsoft 365 workloads are forwarded through your SSE service, and which are not.
    2. Write the bypass decision down before you need it. Microsoft’s own workaround in this incident was to disable GSA. Decide who may approve that, what security controls you give up while it is off, and how you turn it back on.
    3. Keep a break-glass path. Have a tested way to reach critical services if the edge is degraded, such as a small group of pre-approved devices or a documented exception policy, with logging and a time limit.
    4. Test fallback policies in pilot rings. Try switching a small group off the edge during a planned test, so the first time you do it is not mid-incident.
    5. Monitor from the user’s side. Synthetic checks that sign in and open mail or files from the regions you care about will catch a path problem that the provider’s dashboard may not describe yet.
    6. Add the SSE service to your incident triage. When several Microsoft 365 apps fail at once for one region, check service health for the edge service early.
    7. Ask for the post-incident report. Read it, and ask your account team what changes are planned.

    The Bottom Line

    This was a real incident of nearly eleven hours, with a named cause in Microsoft’s Global Secure Access infrastructure, and a scope that narrowed from Germany and the UK to Germany. It was not a global Microsoft 365 failure. The planning lesson, in my view, is still worth taking: if a security service sits in front of your productivity suite, write the runbook for its bad day before it has one.

  • Ivanti Zero-Days Gave Attackers Root on a VPN That Was Fully Patched at the Time: Why Your Remote Access Appliance Needs a Compromise Plan, Not Just a Patch Plan

    Ivanti Zero-Days Gave Attackers Root on a VPN That Was Fully Patched at the Time: Why Your Remote Access Appliance Needs a Compromise Plan, Not Just a Patch Plan

    Patching on day one is the standard advice for edge devices. This week’s Ivanti news shows the limit of that advice. Volexity reports that an attacker used a chain of three zero-days against an appliance that was fully patched at the time, and the lesson for Ivanti Connect Secure zero-day VPN security is that a patch plan alone is not enough. You also need a plan for the day you find out the box was already owned.

    What Volexity Found

    In its October 13, 2026 report, Volexity says that in the first week of October it “discovered an unknown threat actor, UTA0576, abusing a zero-day exploit chain to compromise a fully patched (at the time, version 22.8R2.4) Ivanti Connect Secure (ICS) VPN appliance.” The chain “combined an authentication bypass, a remote code execution (RCE), and a privilege escalation vulnerability to achieve unauthenticated root-level code execution.”

    Two details matter. First, “fully patched” describes that one appliance at that moment. Volexity describes a single investigation at one customer, and I am not claiming this was a broad campaign. Second, Volexity says exploitation began on October 3, 2026, and that it started its investigation on October 5.

    After getting in, Volexity says UTA0576 “primarily focused on credential theft, data exfiltration, log tampering, and network reconnaissance.” It lists:

    • Extracting credentials stored on the appliance, “such as domain credentials used with LDAP.”
    • Searching Active Directory for LAPS local administrator passwords and BitLocker recovery keys.
    • Password spraying and packet capture of LDAP traffic from the appliance.
    • Removing entries from the appliance’s debug log and Linux audit log.

    SecurityWeek adds Volexity’s view that the actions pointed to “preparation for lateral movement and a wider compromise of the network.” Volexity found no backdoor or web shell, because, as Volexity puts it, it “did not identify first-stage malware or web shells that would have enabled the threat actor to execute commands without re-exploiting” the appliance. The entire chain ran for each command, with results returned in the HTTP response.

    That is why detection is hard. Volexity says Ivanti’s Integrity Checker Tool “does not detect any modifications” because the exploit patches code in memory and restores configuration data afterward. “As such, forensic investigation is required to identify evidence of exploitation.”

    Ivanti Connect Secure Zero-Day VPN Security: What Is Patched as of October 15

    Ivanti is tracking the authentication bypass and the RCE under CVE-2026-104935. SecurityWeek reports those are patched in ICS versions 22.8R2.5, 25.1.1.2, 25.1.3.0 and 25.1.3.1. Ivanti’s own October Patch Tuesday post lists the same fixed builds and says Ivanti is “aware of limited exploitation.”

    The third flaw is a privilege escalation in how the dsinvoked service trusts DSCache configuration data. Volexity says “Ivanti is working on a fix for the privilege escalation vulnerability, and it will be included in a future release in the coming weeks.” SecurityWeek says it “remains unpatched.” Ivanti’s post describes “a fix for a related privilege escalation flaw used in the same chain” as “due in the coming weeks.” I checked again today, October 15, and found no report of a released fix. I could not read Ivanti’s advisory page itself, which is a script-loaded page, so that status rests on Volexity, SecurityWeek and Ivanti’s blog. Verify it on Ivanti’s site before you act.

    Do not tell your leadership “all three are patched.” Two are. The third is, by these sources, still open.

    What CISA Says

    In its October 13 alert, CISA says it “added CVE-2026-104935 to its Known Exploited Vulnerabilities (KEV) Catalog on Oct. 13, 2026,” and that “Organizations seeking to determine whether they were compromised will need to conduct network and memory forensics.” The KEV entry in CISA’s catalog data lists a due date of October 16, 2026. For suspected compromise, CISA recommends actions from Ivanti’s recovery procedures: back up the configuration, “conduct a factory reset and upgrade the appliance to the latest version” (rebuild virtual instances), restore the configuration, and “reset passwords for all accounts, including domain, administrator, and local accounts,” plus revoke and reissue stored certificates and reset stored API keys.

    CISA also notes that “Ivanti Connect Secure version 22.8 reaches end of life on Jan. 31, 2027,” which Ivanti’s post echoes while recommending a move to 25.x.

    A note on other vendors. Ivanti’s Patch Tuesday post says edge devices “dominated the weeks before this Patch Tuesday,” citing a Citrix NetScaler zero-day and a SonicWall SMA1000 flaw. SecurityWeek separately reports exploitation attempts against SonicWall’s CVE-2026-102255 starting October 8, after an October 6 disclosure. None of those sources ties these cases to the Ivanti activity, so treat them as a pattern of edge-device attention, not one campaign.

    My Compromise Plan for a Remote Access Appliance

    This checklist is my own recommendation. The sources above do not prescribe it, apart from the CISA recovery steps already quoted.

    1. Patch now, then assume nothing. Apply the fixed build for your code line. Treat the patch as closing the door, not as proof nobody came through it.
    2. Hunt back to early October. Volexity suggests watching for large numbers of requests to welcome.cgi and for mismatches between the URL path and the X-DSCGI-PATH header, and checking /tmp for post-exploitation files.
    3. Look for movement from the appliance. Volexity says systems showing “attempts to move laterally originating from the ICS VPN device” should be forensically investigated. Review directory service queries and logins from the appliance’s service account.
    4. Plan the credential reset before you need it. The appliance can hold domain credentials. Know which service accounts it uses and how you would rotate them, along with certificates and keys.
    5. Limit what the box can reach. Give its service account least privilege and put internal segmentation behind it, so root on the appliance is not a master key.
    6. Keep logs off the box. Volexity says the attacker edited local logs. Forward logs and network telemetry elsewhere.
    7. Rehearse a rebuild. Know who can factory reset and reload the appliance, how long it takes, and what remote access you offer meanwhile.
    8. Know your end-of-life dates. If you run 22.8, plan the move to 25.x now.

    The Takeaway

    This is my view: a patch plan answers “are we on the latest version?” A compromise plan answers “if this box were owned last week, would we know, and what would we do?” For the device sitting at your network edge, you need both, and this week’s reporting shows why.

  • An Executive at a Ransomware Negotiation Firm Was Arrested in the ShinyHunters Probe: How to Vet the Incident Response Firms You’d Call on Your Worst Day

    An Executive at a Ransomware Negotiation Firm Was Arrested in the ShinyHunters Probe: How to Vet the Incident Response Firms You’d Call on Your Worst Day

    Most companies sign an incident response retainer once, file it, and never look at it again. That retainer names the people who would see your most sensitive data, talk to your attackers, and possibly move your money on the worst day your company has. This week’s news is a good prompt to treat ransomware negotiation firm vetting incident response planning as a trust decision you revisit, not a box you checked once.

    What Was Reported, and What Is Confirmed

    On October 9, FBI Director Kash Patel announced that agents had arrested “another suspected co-conspirator of the ShinyHunters group,” as BleepingComputer reported. Patel did not name the suspect. ShinyHunters is the extortion group the FBI holds responsible for the September breach of its jobs portal.

    KrebsOnSecurity reported that the FBI arrested “an executive at a Canadian cybersecurity firm in connection with an investigation into the ShinyHunters hacking group,” citing multiple sources. Press reports identify the man reportedly arrested as Edward Dubrovsky, 54. The FBI has not publicly confirmed that he is the person Patel described. BleepingComputer says plainly: “While the FBI has not publicly confirmed that Dubrovsky is the suspected ShinyHunters co-conspirator, KrebsOnSecurity reports that multiple sources linked Dubrovsky’s arrest to the ShinyHunters investigation.”

    The court record is more concrete, and it spells the name differently. A federal docket indexed on CourtListener, United States v. Dobrovsky in the Eastern District of Pennsylvania, shows an arrest on October 8. It lists the complaint charges as:

    • “CONSPIRACY TO THREATEN TO IMPAIR THE CONFIDENTIALITY OF INFORMATION WITH THE INTENT TO EXTORT MONEY”
    • “INTERFERENCE WITH COMMERCE BY THREATS (HOBBS ACT EXTORTION AND CONSPIRACY TO COMMIT HOBBS ACT EXTORTION)”

    A commitment order dated October 9 sends the defendant to the Eastern District of Texas, where the charges were filed. The complaint itself is sealed. BleepingComputer notes that “it is still unclear what he is accused of doing or whether the case is connected to the FBI breach.”

    These are allegations. Dubrovsky has not been convicted of anything, and he is presumed innocent. Prosecutors have not said the charges involve his negotiation work, and I am not suggesting they do.

    What the Firms Say

    Descriptions of his role vary, so here is what each source says. KrebsOnSecurity reported that LinkedIn listed him as one of CYPFER’s “ex-founders,” and that he is “now associated with another Canadian security firm and sponsor called CyberSteward.” CyberScoop describes CYPFER as “a firm that helps organizations negotiate with ransomware operators,” and CyberSteward as “a Toronto-based firm that helps breach victims negotiate ransom payments.”

    CYPFER disputes the founder label. A firm representing CYPFER told CyberScoop that his role “was Managing Director from May 2022 through Dec. 2025,” and that he “had no affiliation, operational role or relationship with CYPFER since that time, and contrary to certain reporting, was never a founder or co-founder of the company.” I found no published statement from CyberSteward.

    Did he do negotiation work himself? The sources point that way without proving it. Krebs quotes a LinkedIn post in which he wrote of “advisory, negotiations and settlement services,” and one of Krebs’s sources said “the suspect’s company specialized in handling ransomware negotiations with cybercrime groups.” He also recently published a book, Cyber Extortion Strategic Response, about handling cyber extortion.

    One more line from Krebs is the reason this matters beyond one case. Sources told KrebsOnSecurity that “charges against principals at other companies that specialize in ransomware negotiation may be forthcoming.” That is unconfirmed, and I would treat it that way.

    Why an IR Retainer Is a Trust Relationship

    From here on, this is my opinion and my own recommendation, not something any source above prescribes.

    Think about what you hand an incident response or negotiation firm during a crisis. You give them access to your environment, your stolen data samples, your executives’ phone numbers, your insurer, and sometimes your payment process. You give them direct contact with the people extorting you. Few vendors get that much trust that fast, and most companies picked theirs from an insurer’s panel list years ago.

    Picture this hypothetical. Your company is hit, you call the firm on your retainer, and the negotiator assigned to you works through a channel nobody on your side can see. If that person, or anyone at the firm, were ever compromised or conflicted, you would have little way to know. Nothing in this week’s case shows that happened to any client. The point is that most retainers are not built to detect it if it ever did.

    My Ransomware Negotiation Firm Vetting Checklist for Incident Response

    This is what I would do, and it is my recommendation only:

    1. Re-vet the firm every year, not just at signing. Confirm who owns it, who runs it, and whether anything has changed. Check the names on the retainer against current news and court records.
    2. Know the individuals, not just the logo. Ask which people would actually work your case and negotiate, and what background checks the firm runs on them.
    3. Ask how they prevent conflicts of interest. Ask whether they have policies against staff having any relationship with threat actors beyond case communication, and how they enforce them.
    4. Keep visibility into every attacker conversation. Require that all negotiation messages be logged and shared with your counsel and your internal lead, with no side channels.
    5. Split the money from the talking. Any payment decision and movement of funds should need approval from your own executives and counsel, with documented checks, not just the negotiator’s word.
    6. Have a second option ready. Keep an alternate firm pre-approved with your insurer, so you are not stuck with one vendor if you lose trust mid-incident.
    7. Put audit rights in the contract. Ask for the right to review case records, data handling, and staff access after an incident closes.

    The Bottom Line

    One sealed complaint does not tell us what happened, and the person charged has not been convicted. What it does is remind us that the vendors we would call on our worst day deserve the same scrutiny we give any other privileged access. Review that retainer before you need it.

  • The White House Says AI Companies Must Report Model Incidents: What It Means for Your Own Agent Logs

    The White House Says AI Companies Must Report Model Incidents: What It Means for Your Own Agent Logs

    On October 9, Axios ran a story headlined “Exclusive: Anthropic breaches spark White House AI reporting mandate.” Since then, “AI incident reporting mandate” has been the shorthand for what happened. The real picture is narrower and looser than that phrase suggests. It is still worth an IT team’s attention, mostly because of what it implies about the logs you keep on your own AI agents.

    What the Statement Actually Says

    The text comes from a statement by leaders of the White House Super Intelligence Force (SI Force), which Axios says was “shared exclusively with Axios.” Axios published it in full in its October 9 article. Every quote in this section comes from that statement as Axios published it.

    The core lines:

    • “SI companies must immediately disclose incidents involving their models and follow with swift, decisive action to remedy any and all harm.”
    • “This notification and remediation process is not optional. It is a critical national security obligation.”
    • “We expect Anthropic and all companies to honor their obligations, immediately report incidents, fully cooperate with federal and state law enforcement authorities, remedy any damage, and implement concrete safeguards to ensure these failures do not reoccur.”
    • “Our message to all SI companies is clear: delayed notification, inadequate corrective action, and a failure to take responsibility will not be tolerated.”

    (“SI” is the administration’s term for AI. A September executive order directs agencies to use “Super Intelligence” in place of “Artificial Intelligence.”)

    As for what prompted it, the statement says: “Earlier today, Anthropic contacted the SI Force to disclose the details of various prior incidents that it discovered in late September involving the unauthorized and fraudulent use of government and other systems.” It adds that “The company informed us that these events occurred in the past, the activity has ceased, and there is no ongoing similar activity.” I am leaving the characterization there. The quoted words are the statement’s, not my own findings.

    What It Does Not Say

    This is the part that matters for anyone deciding whether to change process. As of October 13, 2026, I could not find any executive order, OMB memo, or Federal Register rule that puts this requirement in writing. I checked whitehouse.gov’s presidential actions and statements and searched the Federal Register. The only text is the statement Axios published.

    Read closely, the statement also leaves out the pieces a real reporting regime would need:

    • No definition of an incident. It refers only to “incidents involving their models.”
    • No deadline beyond “immediately.” There is no hour count.
    • No reporting channel. It does not say where, how, or to which office a report goes.
    • No penalties. Axios notes that “The statement did not make clear what enforcement mechanisms or penalties would look like if AI companies failed to disclose incidents and remediate them.”

    Scope matters too. Axios reports that “The White House requirements apply to all AI companies,” and the statement speaks to “SI companies” and their models. Nothing in it addresses organizations that simply use AI tools or run agents built on someone else’s model. If you run an IT shop, the statement is not addressed to you.

    Background: The 72-Hour Clause That Came and Went

    Federal AI incident reporting has been tried before, in a narrower form. OMB memo M-24-18 (September 24, 2024) told agencies to, “to the greatest extent practicable, include contractual terms requiring that vendors have a process for identifying and disclosing to agencies serious AI incidents and malfunctions of the acquired AI system, or service within 72 hours, or a timely manner based on the severity of the incident.” It left the definition to agencies: “Criteria for what constitutes a serious AI incident or malfunction should be determined by each agency.”

    That was a contract clause for federal purchases, not a rule for industry. And it did not last. OMB memo M-25-22, dated April 3, 2025, states: “This memorandum rescinds and replaces OMB Memorandum M-24-18.” I searched M-25-22’s full text and did not find a comparable incident-reporting clause.

    Why an AI Incident Reporting Mandate Still Points at Your Agent Logs

    Everything from here down is my own recommendation. The statement does not require any of it, and it does not apply to companies that only use AI.

    Here is why I would act anyway. The message from Washington is that fast disclosure of AI incidents is now expected, at least from model makers. Expectations like that tend to flow downstream through contracts, vendor questionnaires and cyber insurance forms. When that happens, the first question will be simple: what did your agent do, and can you prove it?

    Picture this hypothetical. An internal agent with web access submits a form to an outside site it was never meant to touch. Your vendor calls and asks for a timeline. If your logs show only “agent ran,” you cannot answer. If they show every action, with inputs and approvals, you can answer in an afternoon.

    What I would log for every AI agent you run:

    1. Identity. Which agent, which model and version, and which human or service account it acted for.
    2. Every external action. Outbound requests, form submissions, emails sent, files written and API calls, with timestamps and destinations.
    3. Tool calls and permissions. Which tools it invoked, with what arguments, and whether a human approved them.
    4. Prompts and key outputs. Enough to reconstruct why it did something, stored with the same care as other sensitive logs.
    5. Blocks and failures. Refused actions and guardrail hits are often your earliest warning.

    Then put the logs somewhere useful. Send them to the same SIEM as everything else, retain them long enough to reconstruct a timeline weeks later, and lock them so the agent itself cannot edit or delete them.

    Finally, write the playbook before you need it. Decide now what counts as an agent incident in your shop, who gets told, and how fast. Neither the statement nor any rule defines that for you, so your own definition is the one that will matter.

    The Bottom Line

    What exists today is a strongly worded White House expectation aimed at AI companies, published through one outlet, with no written rule behind it yet. That could change. Either way, being able to say exactly what your agents did is good security practice on its own terms. That is my view, and it is the reason I would start logging now.

  • Attackers Are Exploiting AhsayCBS Flaws to Take Over Backup Servers: Your Recovery Plan Is Now an Attack Surface

    Attackers Are Exploiting AhsayCBS Flaws to Take Over Backup Servers: Your Recovery Plan Is Now an Attack Surface

    Backup servers are supposed to be the boring, trusted box in the corner that saves you on your worst day. This week is a reminder that attackers see them differently. Huntress reports that threat actors are chaining two AhsayCBS flaws to run code on exposed backup management servers, and the vendor’s first patch did not fully close the hole. If you treat AhsayCBS vulnerability backup server security as a quarterly checkbox, now is the time to treat it as an active incident.

    What Huntress Saw in the AhsayCBS Attacks

    AhsayCBS (Cloud Backup Server) is the management console for Ahsay’s backup software. In its October 8, 2026 writeup, Huntress says it is “primarily used by managed service providers (MSPs) and system integrators” and that it “centralizes control of backup operations.” That makes it a high-value spot: one console, many customers’ backups.

    Huntress says NVD disclosed two CVEs on October 4, 2026:

    • CVE-2026-105133, which Huntress describes as “a medium-severity flaw” in the checkSysPwd function that “can lead to improper authentication.”
    • CVE-2026-105134, described as “a critical-severity vulnerability” in the Replication Receiver component that “can be exploited to achieve unauthenticated remote code execution as NT AUTHORITY/SYSTEM on the host.”

    According to Huntress, attackers are chaining them: “First, CVE-2026-105133 is used to bypass authentication. Then CVE-2026-105134 is used to gain code execution.” Huntress says exploitation started on October 7 at 23:20:15 UTC, which is 7:20 PM ET. It adds that “As of October 8, Huntress has seen five organizations targeted via these flaws.”

    Here is what Huntress documented after the attackers got in:

    • Webshells. “In some incidents we saw threat actors deploy .jsp webshells in the web application directory immediately after exploitation.”
    • A disguised cryptominer. “On several endpoints we observed the XMRig miner (edge.exe), renamed to masquerade as a Microsoft Edge process.”
    • Persistence. A Windows service named MicrosoftEdgeUpdateSvc, built to look like the real Edge Update service, which ran a modified copy of the NSSM utility with SYSTEM privileges.
    • Hiding tricks. A PowerShell script, Taskgmr.ps1, that Huntress says “appears to be an AI-assisted script.” It stopped the miner’s service while Task Manager was open.
    • A vulnerable driver. In one incident, attackers downloaded WinRing0x64.sys. Huntress says it gave the miner kernel-level access to the hardware.

    Huntress does not name an attacker, and its report does not describe any data theft or tampering with backups. I am keeping that scope. What it documents is SYSTEM-level code execution, webshells and mining. That is bad enough on a box that controls backups.

    The Patch Story: Incomplete, Then Corrected

    I would not call these flaws unpatched. A fix exists. But the timeline has a twist every operator should know. Ahsay’s own October 10, 2026 security alert says: “our previous patch v10.3.4.0 did not fully address the following security vulnerabilities,” naming both CVEs. It tells partners that “Partners running affected v10 versions should apply the AhsayCBS v10.3.4.45 hotfix” and adds: “After applying the hotfix, please reboot your server.” Ahsay says a partner portal login is required to download it.

    Huntress’s own updates track the same correction. On October 8 it said 10.3.4 “is also affected,” noting that “Previous reporting indicated that 10.3.4 was not susceptible.” Its October 12 update says it “independently verified that the AhsayCBS 10.3.4.45 hotfix mitigates the initial entry point of this attack chain.” Huntress also says companies on AhsayCBS v7, v8 and v9 “should apply the 9.15.2.70 hotfix.” Ahsay’s alert only mentions v10, so check your version against both pages.

    The lesson there is simple. If you patched to 10.3.4.0 and moved on, you are not done.

    AhsayCBS Vulnerability Backup Server Security: What Operators Should Do Now

    Huntress recommends restricting management interface access “to trusted IP addresses only or require VPN.” If you find any of its indicators, it says to “carry out a full host re-image from a trusted backup; as attackers have been able to hide secondary backdoors for extended persistence.” My checklist, built on those two sources plus my own judgment:

    1. Apply the hotfix and reboot. For v10, that is 10.3.4.45 per Ahsay. Confirm the reboot actually happened.
    2. Get the console off the open internet. Allow-list trusted IPs or put it behind a VPN, as Huntress advises.
    3. Hunt before you relax. Huntress publishes indicators and four Sigma rules. Look for unexpected child processes from cbssvcX64.exe, .jsp files in the web app directory, a MicrosoftEdgeUpdateSvc service, Edge-named binaries in Temp folders, WinRing0x64.sys, and outbound traffic to mining pools on port 8029.
    4. Re-image if you find a hit. Do not just delete the webshell. Huntress’s guidance is a full rebuild from a trusted backup.
    5. Review downstream access. If the console has credentials or agents tied to customer environments, check whether any of those were used during the exposure window.

    Opinion: Your Recovery Plan Is Now an Attack Surface

    This section is my opinion, not a finding from Huntress or Ahsay. In this campaign, the documented payoff was cryptomining. But think about what SYSTEM access to a backup console could mean in a different campaign. Imagine an attacker who quietly changed retention policies or disabled jobs weeks before a ransomware event. You would not find out until you tried to restore. That scenario is hypothetical, but the access needed to attempt it is exactly what these flaws grant.

    So I treat backup management the way I treat domain controllers. Keep it off the internet. Give it separate admin credentials with MFA. Monitor it like a crown-jewel system. Test restores from copies the console cannot delete or change. If your recovery plan depends on a box anyone can reach from the internet, the plan is part of your attack surface.

  • Data Centers Are Waiting in Line for Power: What Grid Interconnection Delays and Temporary Local Moratoriums Mean for Your Capacity Plan

    Data Centers Are Waiting in Line for Power: What Grid Interconnection Delays and Temporary Local Moratoriums Mean for Your Capacity Plan

    Imagine your 2027 capacity plan assumes a new colo cage in a fast-growing market, or a cloud region expansion that your provider has hinted at. Now imagine that plan depends on two things that never show up in a hardware quote: a utility’s interconnection study and a county board vote. That is the planning risk behind data center power grid interconnection delays, and the 2026 evidence says it deserves a line on the risk register.

    I want to be careful about scope here. I am not going to claim the chip shortage is over, or that power is now the only constraint. No source I checked says that. What the sources do show is that grid operators are rationing how fast large loads connect, and that some local governments have paused new data center approvals while they study power and water impacts.

    Texas: A Queue Bigger Than the Grid

    ERCOT, the Texas grid operator, published a June 18, 2026 explainer on its new batch connection process. It says that as of mid-2026, large users “totaling more than 438,000 MW of proposed demand had entered ERCOT’s connection queue,” and that “Nearly 90% of those requests were from data centers.” For scale, ERCOT notes its all-time peak demand record is 85,508 MW, set on August 10, 2023. It defines a large user as a single site with peak demand above 75 MW. A later ERCOT presentation to the Texas Senate on July 29 put the figure at approximately 474 GW of large loads seeking interconnection as of June 2026, with about 90% data centers.

    Those are requests, not buildings. ERCOT says plainly that “not all connection requests result in built projects.” The interesting part is how it is responding. ERCOT says studying requests one at a time “triggered costly restudies that delayed completion timelines.” So it moved to a batch process designed so large users “only connect in quantities and locations the Texas transmission grid can support.” Under that process, a project may get its full requested capacity, “a partial allocation, or may need to wait until a subsequent batch.”

    ERCOT’s own published timeline shows how far out this runs. Batch Zero began in July 2026. Each large user gets its reliable allocation in spring 2027, a final transmission plan follows in fall 2027, and applications for Batch 1 open in summer 2027. ERCOT also offers optional pathways for customers that bring on-site generation or agree to let ERCOT curtail their load during local grid stress.

    California and Florida: Temporary Local Pauses

    The moratorium part of the headline holds up, with an important qualifier: these are temporary, local land-use pauses, not statewide bans.

    • San Joaquin County, California. In an October 7, 2026 release, the county said its Board of Supervisors approved an interim urgency ordinance “establishing a 45-day moratorium on land use approvals for new data center development in unincorporated San Joaquin County.” The county cited data centers’ need for “substantial and continuous electricity and water resources,” and said it “must determine whether existing electrical infrastructure can accommodate multiple large-scale data centers.” It also says there are currently no large-scale AI data center proposals in unincorporated parts of the county, so this is a pause on future approvals, not a blocked project.
    • Manatee County, Florida. A county government release published by The Bradenton Times on October 8, 2026 says the commission approved “a temporary, 12-month moratorium on new data center-related development applications in unincorporated Manatee County.” The stated purpose is time to evaluate how data centers and other large-load facilities fit the county’s plans and regulations. The release also notes that large-scale data centers “can require substantial amounts of electricity and water to support continuous operations and cooling systems.” It also says the moratorium “does not mean data centers have been permanently prohibited.”

    On the utility side in California, the picture is more mixed than “everyone is stuck.” PG&E’s own October 5, 2026 newsroom piece says that “Across the country, the time needed to get power is slowing data center projects.” It then argues its own grid “has open capacity today” and that some parts of its territory can add 50 to 100 MW using existing equipment. It adds that this “does not mean every project can connect at once or in any location.” That is a utility describing its own system, so I would treat it as a positioning statement worth verifying site by site, not a guarantee.

    What Data Center Power Grid Interconnection Delays Mean for IT Capacity Planning

    The rest of this post is my opinion and practical advice, not a finding from the sources. Most IT shops do not file interconnection requests. Our exposure is indirect: the colo provider, the cloud region, or the hosting partner that does. But indirect is not the same as absent. If a provider’s new hall depends on a batch allocation that lands in spring 2027, your expansion date depends on it too.

    A few things I would do now:

    • Ask providers where the power is, not just where the space is. For any new site or expansion on your roadmap, ask whether its utility service is energized, contracted, or still being studied. “Coming soon” is not a power status.
    • Add lead-time buffers for new markets. Treat capacity in a market with a crowded queue or an active local pause as later and less certain than the sales deck says.
    • Watch the county calendar. Local pauses in 2026 have been temporary and tied to study periods. They can be extended, lifted, or turned into new zoning rules. If a site you depend on sits in an unincorporated county, someone should be tracking those hearings.
    • Keep a second option warm. Identify an alternate region, provider, or existing site that already has energized capacity, so a slip in one market becomes a migration plan instead of a missed launch.
    • Separate demand you control from demand you hope for. ERCOT plans a commitment deadline for Batch Zero developers to post financial security and prove site control. Separating firm from hopeful demand is a decent discipline for internal forecasts too.

    The Plan Now Has a Utility Dependency

    For years, capacity planning was mostly about budget, space and hardware lead times. Those still matter. The 2026 record adds two more dependencies most of us did not model: a grid operator deciding how much load a region can take, and a local board deciding whether to approve new facilities at all. Neither one is a reason to panic. Both are reasons to ask better questions before you sign.

  • Anthropic’s Cyber Verification Program Reduces Claude Security Blocks for Verified Defenders: Who on Your Team Should Hold the Badge

    Anthropic’s Cyber Verification Program Reduces Claude Security Blocks for Verified Defenders: Who on Your Team Should Hold the Badge

    Picture this: your building hands out a new master key, and only people who pass a background check can get one. That sounds sensible until you are the one deciding which of your employees carries it, which doors it opens, and who reviews the log afterward. That is roughly the governance question Anthropic’s Cyber Verification Program now puts in front of security teams. The program reduces some Claude security blocks for verified defenders, and it hands each organization the job of deciding who internally gets the sharper tool.

    Here is what Anthropic actually documents, what it does not, and the questions I would want answered before turning it on.

    What the Anthropic Cyber Verification Program Changes for Claude Security Work

    Anthropic’s October 6, 2026 announcement describes an “expanded version” of the Cyber Verification Program, or CVP. It says the CVP “makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals.” The starting point matters. Anthropic says its generally available models, such as Claude Opus 5.5, Claude Fable 5.1 and Claude Sonnet 5.5, “have conservative cyber safeguards that block most cyber work.” Its reasoning is that cybersecurity is “inherently dual use”: the same capability that helps a team find and fix a flaw can help an attacker exploit it.

    The general models are not useless for security. Anthropic’s Help Center page for the program says they can still be used for secure code review, threat modeling, patching known issues, finding vulnerabilities in your own source code and triaging alerts. But it warns that “other cyber security work like malware analysis or exploit validation may be interrupted by our safety classifiers.” The CVP is the route for professionals whose legitimate work is getting blocked.

    This is also a merger. Anthropic says that for the past six months it ran two programs: Project Glasswing, which gave organizations securing critical software access to Claude Mythos, and the original CVP, which gave vetted teams “access to reduced safeguards on Claude Opus and Claude Sonnet models.” The new CVP integrates both. Each tier includes Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1 and future models.

    Three Tiers, Documented

    Anthropic documents three access tiers, each with “different verification requirements and security controls”:

    • Defense Access covers defensive work such as SOC and incident response tasks, reverse-engineering malware, and analyzing and validating vulnerabilities. Anthropic’s examples of qualifying organizations include corporate, nonprofit, university and government security teams defending their own systems; critical infrastructure operators of any size; smaller security firms; open-source maintainers; and individual researchers with a track record of reported vulnerabilities. It is the only tier open to individuals, and only on a paid plan.
    • Red Team Access adds authorized penetration testing and red-teaming, for organizations only. Testing is limited to systems they are authorized to test. Anthropic says users “will still experience real-time blocks” on actions that could cause physical harm or mass disruption, such as deploying ransomware or damaging physical systems.
    • Specialized Access “has the fewest cyber blocks.” It is reserved for a limited set of verified organizations authorized to test safety systems such as flight operating systems, power grids, telecom networks and interbank transfer infrastructure. Anthropic says it reviews every one of them “in depth in collaboration with the US government,” and existing Glasswing members move into this tier.

    Anthropic also published results from its own evaluation, CyScenarioBench, run against Claude Opus 5.5 under each tier’s safeguards. It reported that without CVP every task was blocked on the first prompt, and that in Defense Access 46 of 50 trials were blocked at some point. In Red Team Access there were no blocks, and the model completed 34 of 50 tasks. That is vendor-run testing on a vendor-built benchmark, so read it as Anthropic’s evidence that the tiers behave differently, not as independent validation.

    Who Checks the Badge

    Here the documentation is useful but not complete. Anthropic says it “will verify all applicants and request proof of the required security controls for the relevant access tier.” The Help Center lists what an application needs: organization and applicant details for identity verification, a description of your security work, and “an attestation to the security controls applicable to your desired tier.” Organizations apply once, and admins then designate seats or assign the program to workspaces or custom roles. Anthropic aims to answer within seven business days and places each applicant “at the highest tier” its information supports. The announcement adds that Red Team reviews take “a few weeks.”

    The Help Center also lists factors that can rule an organization out. These include the nature of the work, whether Anthropic can verify who the organization is, the legal and regulatory environment and the risk access “could be compelled,” and a more cautious approach for organizations that primarily serve military, intelligence or law enforcement customers.

    A few strings stay attached. The Usage Policy “still applies in full.” Anthropic says it “may review, narrow, or withdraw a grant.” Data retention is required “so that we can monitor for cyber misuse,” with limited zero-data-retention exceptions until Anthropic’s Enterprise Frontier Safeguards option arrives. Defense Access members have until December 15, 2026 to adopt phishing-resistant MFA and stop using API keys. CVP access also cannot be used to build a client-facing product without a separate approval.

    What is not public: the exact scoring behind tier placement, how attestations are checked in practice, and precisely which classifier behaviors change in each tier beyond Anthropic’s examples. SecurityWeek’s coverage matches Anthropic’s description of the three-tier structure, but I have not seen independent detail beyond that.

    Opinion: The Real Badge Check Happens Inside Your Org

    This section is my opinion. Anthropic is checking the organization. Once a grant lands, your admins decide which workspaces, roles and seats get it. That makes CVP an internal privileged-access problem as much as a vendor program, and I would govern it like one:

    • Map the tier to job duties. Only people whose work actually needs malware analysis or exploit validation should sit in a CVP-enabled workspace.
    • Treat the workspace like a privileged group. Use named seats, periodic access reviews, and removal on role change.
    • Log on your side too. Anthropic retains data for misuse monitoring, but that is its monitoring, not yours. Keep your own record of who used the elevated workspace and for which engagement.
    • Get the controls done early. Phishing-resistant MFA and moving off API keys are good hygiene anyway, and the deadline is documented.
    • Write down scope. Red Team Access is limited to systems you are authorized to test. Keep that authorization paperwork next to the access list.

    A sharper tool for defenders is a reasonable idea. The part that will decide whether it stays reasonable is the boring part: who holds the badge, and who reads the log.

  • IDC Frontier Unauthorized Access Causes a Partial Cloud Outage for 495 Customers: Why You Need an Exit Plan for Your Regional Host

    IDC Frontier Unauthorized Access Causes a Partial Cloud Outage for 495 Customers: Why You Need an Exit Plan for Your Regional Host

    Imagine you run IT for a mid-sized town. Your disaster recovery plan has a tidy section on what happens if a big hyperscaler region goes dark. It probably has nothing on what happens if the quiet regional host that runs your website and a couple of line-of-business apps gets breached, and then has to pull the plug on its own platform to contain it. That second scenario is roughly what the IDC Frontier unauthorized access incident looks like from the customer side, and it turned into a partial cloud outage that operators everywhere should study.

    Here is what the provider has actually said, what it has not, and what I would do with it.

    What IDC Frontier Says Happened

    IDC Frontier runs IDCF Cloud, and its own notices describe the parent company as SoftBank Corp. In its second notice on October 7, 2026, the company said that since about 3:40 a.m. Japan time a fault had been occurring in East Japan Region 1. It said its investigation had found the fault was caused by “a ransomware attack by a third party.” It listed the scope as “495 companies and local governments that have contracted IDCF Cloud,” and said it was contacting affected customers individually.

    That 495 figure is one combined count of contracted customers, companies and local governments together. IDC Frontier has not published a split between the two, so I am not going to invent one.

    The response is the part operators should read twice. IDC Frontier said that “to prevent secondary damage and data leakage,” it cut East Japan Region 1 off from the network and then stopped the systems. As a precaution, it also stopped the customer-facing management consoles for its other regions while it checked whether they were safe, with plans to restore them once that was confirmed.

    How Unauthorized Access Became a Partial Cloud Outage

    The third notice, on October 8, narrowed the damage. The affected area is part of East Japan Region 1, specifically the tesla, henry, pascal and joule zones. There, virtual servers have stopped and cannot be restarted. IDC Frontier said customer data in those four zones is expected to be “difficult to extract or restore.” In its current view, recovery is possible only from backups customers hold themselves, and it is advising those customers to prepare a separate environment and rebuild.

    For the remaining zones of East Japan Region 1 (radian and newton), East Japan Regions 2 and 3, and West Japan Region 1, the company said no unauthorized access has been confirmed so far. It is still asking those customers to take their own backups. Because the consoles are down, IDC Frontier staff are starting and stopping customers’ virtual servers on request. The notices also say IDCF Cloud TypeS and IDCF Private Cloud are outside the scope of the incident.

    So the outage was partial, and it has two layers. IDC Frontier attributes the stopped, unrestartable servers in four zones to the ransomware attack itself. Its own containment then cut the region off the network and took away the self-service consoles everywhere else. Customers in zones with no confirmed intrusion still lost their control panel. I think that distinction matters more than the headline number.

    The downstream effects were visible fast. The Asahi Shimbun reported that the websites of the Ibaraki Prefecture government and prefectural police became inaccessible. It also reported that a logistics subsidiary of seafood company Nissui, which uses an IDC Frontier data center, had its logistics system shut down.

    Who Did It, and What We Still Do Not Know

    IDC Frontier’s notices describe the attacker only as “a third party.” They do not name a group, and neither will I. The company says it is working with an outside security firm on the intrusion route, has reported the incident to supervisory authorities and the Tokyo Metropolitan Police Department, and is still investigating the scope of the impact. Its fourth notice, on October 9, says it set up an emergency task force at 9 a.m. on October 7, working closely with SoftBank. SoftBank’s corporate division is handling customer contact, backup guidance and migration proposals.

    Still open: how the attacker got in, whether any data left the building, and when, or whether, the four zones come back.

    Opinion: The Regional Host Is a Concentration Risk You Can Measure

    This section is my opinion, not a finding from any source. We spend a lot of tabletop time on “us-east-1 is down.” That is a fine exercise, but it trains us to think of provider failure as an availability event that the provider fixes and we wait out. This incident is a different shape. A provider that is breached may decide, correctly, that the safest move is to isolate its own platform and lock customers out of the control plane. When that happens, your recovery time is set by your own backups and your own rebuild capability, not by the provider’s status page.

    Calling the regional provider “your single point of failure” is my framing. Whether it is true for you depends on how many of your services sit there and what you can stand up somewhere else.

    What I Would Check About My Own Hosting Provider

    • Count what lives there. List every public site, app and integration on each hosting provider, including the ones a vendor runs for you.
    • Own a backup the provider cannot touch. IDC Frontier’s own guidance is that recovery for the four zones depends on customer-held backups. Make sure yours are off-platform and tested.
    • Plan for losing the console, not just the servers. If you could not log in to your provider’s portal tomorrow, could you still fail over DNS, certificates and traffic?
    • Write the exit runbook now. Pick a second environment, document the rebuild steps, and time a restore. “Prepare a separate environment and rebuild” is a bad instruction to receive for the first time in the middle of an incident.
    • Ask the provider the hard questions. How are management plane and customer workloads separated? What would it isolate in a breach, and how would you be told?

    None of this requires distrusting a regional host. Regional providers do a lot of essential work well. It just requires admitting that “the provider will handle it” is a dependency, and dependencies belong on the risk register.