Lead · Vulnerabilities

Two more Ivanti Connect Secure zero-days land in KEV, about a month after someone else found them

Ivanti patched CVE-2026-111357 and CVE-2026-118527 yesterday; the chain has been running in attacks since at least 1 October, which means the disclosure timeline is really an attacker's receipt.

confidence 0.93

Ivanti released updates for two more Connect Secure remote access gateway flaws, and CISA added CVE-2026-111357 and CVE-2026-118527 to the KEV catalogue, after the pair were chained in zero-day attacks since at least 1 October. Read that sentence in the order the calendar demands and it says something slightly different from the press release: for about a month, a perimeter box whose entire job is to be the perimeter was a staging area, and the only people with a full inventory of affected organisations were the ones doing it.

This is the part of the ritual where someone says the word sophisticated. It is not a useful word here. What happened is that an internet-facing appliance with a web interface was found to have two bugs that combine, and the people who found them first were not the vendor and were not us. The patch is the end of the vendor's problem and the beginning of yours, because a device that was exploitable for a month is not a device you patch, it is a device you assume was visited and then go and prove otherwise with logs you hopefully kept.

The NCSC's chief chose the same week to warn of an unprecedented rise in attacks in the coming months, with the line that vulnerabilities tolerated for far too long can now be discovered and exploited at a scale that was once unimaginable. Tolerated is doing enormous work in that sentence, and it is the right word. Nobody accepted this risk in a meeting. We just kept not getting round to it, which is the same thing with better paperwork.

Why it mattersIf you run ICS gateways, patching today is the easy half; hunting the 1 October to 2 November window is the half that tells you whether you have an incident.

Today’s revelation

Every advisory contains two dates, and the one the attacker used is always the older one.

— Sturgeon’s Revelations · Probably Fine Daily No. 37

§2

Top stories

The rest of the last 24 hours, with receipts.

Vulnerabilities

A thousand GitLab servers are still internet-exposed to a maximum-severity flaw that is being exploited now

Over 1,000 internet-exposed GitLab servers remain unpatched against a maximum-severity flaw actively exploited for remote code execution. In the same window Synology told NAS owners to update immediately because of a surge in C3RB3R ransomware attacks against their devices. Neither of these is a mystery, a nation-state, or a novel technique; both are a box somebody stood up to be useful, on the internet, with a version number from before the bad news.

Why it mattersA GitLab box holds your source, your CI tokens and your deploy keys, which makes it less a server and more a skeleton key with a web interface.

confidence 0.90

Vulnerabilities

Roughly a hundred CVEs arrived in one morning, nearly all WordPress, and then NVD went quiet

The feed took about a hundred NVD entries on 31 October, overwhelmingly WordPress plugin bugs in the same two or three flavours: unauthenticated stored cross-site scripting through comment fields, and missing authorisation on endpoints that cheerfully accept whatever you send. Buried among the plugin churn is CVE-2026-119977, remote code execution in WordPress Core versions 4.7 through 7.1.3, caused by predictable web-accessible staging paths during full-package core upgrades, exploitable by unauthenticated attackers during the upgrade window. Then NVD stopped, with nothing after 21:05 on 1 November, which is either a publication pause or an ingestion gap and is not reassuring in either reading.

Why it mattersThe vulnerable moment in CVE-2026-119977 is the upgrade itself, so the one maintenance task you were praised for is also the window.

confidence 0.75

Ransomware

The Gentlemen published about two dozen victims in a morning, and N0n is numbering them like invoices

The Gentlemen posted 24 leak-site items on 2 November against a median of zero on other days, a spread running from a Scottish property law firm and a French logistics holding to a Brazilian state health department and Mitsubishi Motors Ukraine; our tempo count puts the group at 50 claims in seven days against 24.97 expected from its own prior rate, p below 0.001, in a week where 217 claims came from 39 of 269 tracked groups. N0n, separately, is publishing anonymised numbered entries — Company #113 through #122, marked PENDING EXPOSURE, described only by sector, from dental insurance to a UAE retail conglomerate — which is extortion run as a release pipeline with names withheld as the leverage. Everstake has gone further and put countdowns on it: 582.69 Gb scheduled for 10 November, 473.05 Gb for 12 November, victims still redacted. Alongside, Interlock posted two US school districts, Redact claimed 29.7 GB and 11,156 files from law firm Holland & Knight, and SilentRansomGroup is surging at 11 claims against 2.55 expected.

Why it mattersA bulk dump distorts your trend lines: this is a publication event, not twenty-four intrusions that happened on Monday, and your board slide will not know the difference unless you tell it.

confidence 0.85

Threat actors

They phoned as IT support on Teams, got remote control, and asked Active Directory for a certificate

In two human-operated intrusions, attackers impersonated IT help desk staff over Microsoft Teams messages and calls, talked employees into granting remote access through Quick Assist, then abused insecure Active Directory Certificate Services configurations along ESC1 and ESC8 paths to issue login certificates for privileged accounts, including one for a domain controller. No exploit, no malware worth the name at the entry point, just a convincing colleague and a certificate authority that does what it is told. The certificate is the point: it is durable, it is legitimate, and it survives the password reset you will perform while feeling productive.

Why it mattersAudit your AD CS templates for enrollee-supplied subject names and web enrolment over HTTP, and treat external Teams contact and Quick Assist sessions as an authentication event, because that is what the attacker treated them as.

confidence 0.85

AI & security

The models are now on both sides of the invoice: ARTEX hit 200 organisations, and the AI tooling itself got breached

Chinese operators used ARTEX, an open-source multi-agent AI penetration-testing framework, in attacks on more than 200 organisations across 14 countries, with the VAPOR PANDA campaign reported running parallel intrusions from 17 October using DeepSeek and commercial models; a separate open-source framework, CyberStrikeAI, was used against educational institutions in Africa and Latin America; and the VOLTA infostealer calls out to Claude, Gemini and Kimi to rank stolen data by how much the victim is worth. Pointing the other way, Cursor's extension marketplace served malicious Claude Code extensions, versions 2.1.312 and 2.1.313, on 29 October, performing host reconnaissance and attempting to plant SSH backdoors on developer machines; Chainalysis disclosed a breach in which attackers used offensive agentic techniques against its systems and internal tools; and at least nine AI developer platforms were breached during October with credential theft and abuse of model access. Unit 42 also reported SysGoRAT against government Linux infrastructure with suspected AI-generated scripts. The common thread is not intelligence, it is throughput.

Why it mattersYour developer laptops now have an extension marketplace with the blast radius of a package manager, and nobody owns it.

confidence 0.72

Malware & IOCs

Google dismantled a 1.3 million-device botnet and says the manufacturer built the backdoor on purpose

Google and industry partners took down the XCPE botnet, more than 1.3 million Android devices, mostly TV set-top boxes, which rented out owners' internet connections to strangers and ran advertising fraud. Google says it is highly confident that device maker Videostrong intentionally put the backdoor in the boxes and sold access to criminals, and has banned the manufacturer. That is the shift worth noting: not a supply chain compromised by an outsider, but a supply chain working exactly as its owner intended, with the customer as the product being shipped.

Why it mattersCheap consumer boxes on staff home networks are now a vendor-trust problem rather than a patching problem, and you cannot patch a business model.

confidence 0.85

Threat actors

The FBI pins the $6.43 billion Binance theft on TraderTraitor while three more crypto incidents run in parallel

The FBI attributed the theft of more than $6 billion in virtual assets from Binance to TraderTraitor, the North Korea-aligned group also tracked as Jade Sleet, Pressure Chollima, PUKCHONG, Slow Pisces and UNC4899, and has flagged over 600 addresses as the proceeds are laundered. In the same day's feed and with no evidence of any connection between them: Brazil's largest exchange, Mercado Bitcoin, confirmed a cyberattack with more than $46 million leaving the platform; BTCPay Server warned that hot wallets on outdated instances are being drained; and Polymarket is returning $3.9 million to traders after DebugBear's DNS records were compromised via its domain registrar. The Binance post-mortem reporting makes the uncomfortable point plainly: one compromised laptop, weak system separation, systemic loss.

Why it mattersRegistrar and third-party script compromise sit outside most asset inventories and inside every customer's browser session.

confidence 0.80

Oddities

Somebody changed the temperature settings on supermarket freezers, remotely, across several countries

A cyberattack on refrigeration systems disrupted stores run by multiple supermarket chains across several European countries, after attackers remotely changed temperature settings on freezers and cooling systems. There is no CVE in the story, no ransom note mentioned, and nothing clever: a control system that accepted a setpoint from whoever asked. Elsewhere in the strange column, Russia's federal statistics agency Rosstat confirmed a cyberattack claimed by an alliance of Belarusian and pro-Ukrainian hackers.

Why it mattersIf a setpoint can be written remotely without strong authentication, then your environmental control system is an actuator with a login page, and it belongs in the same asset register as the domain controller.

confidence 0.80
§3

Hidden

The daily deep read: what is under the feeds, not on top of them.

Under the feeds: two trackers copying each other, one burst mistaken for a crime wave, and an indicator count that measures plumbing rather than infections.

  1. Exception 01

    Multi-source does not mean corroborated

    ransomware.live consistently precedes ransomlook by about one to one and a half hours on N0n, Everstake and safepay postings, and by four and a half hours in the one orova case where ransomlook came first. The multi-source victim counts for Qilin, Akira and N0n largely reflect shared scraping of the same leak sites, not independent confirmation.

    confidence 0.70
    Wrong if
    Evidence the trackers scrape independently, such as differing victim sets or lead-lag order that varies by group.
    Watch
    Whether the roughly one-hour lag holds, and whether Everstake's scheduled reveal dates of 10 and 12 November actually arrive.
  2. Exception 02

    The Gentlemen's day is a publication, not an intrusion rate

    24 items in roughly three and a half hours on 2 November against a zero median on other days, spanning healthcare, legal, manufacturing, education and automotive across Europe, the Americas, Africa and Oceania, looks like a batch release of a backlog or an affiliate campaign. Per-victim counts derived from it will overstate real-time compromise.

    confidence 0.70
    Wrong if
    Victim disclosures dating the intrusions to the same days, or the group sustaining the same rate for a week.
    Watch
    Cadence over the next three to seven days, and any confirmation from the named organisations.
  3. Exception 03

    SalatStealer's 2,221 indicators are a feed artefact until proven otherwise

    SalatStealer produced 2,221 indicators in about 23 hours against 5 to 52 for every other family in the window, while Mirai and Mozi begin at the same minute and SilverFox and ValleyRAT are identical in count and timing. That pattern points at batch ingestion and aliasing rather than an infection surge.

    confidence 0.50Speculative
    Wrong if
    Independent telemetry showing an actual SalatStealer campaign spike.
    Watch
    Whether the indicator rate stays elevated once the batch clears.
  4. Exception 04

    Education is the sector of the week, by accumulation rather than campaign

    BYOD's unconfirmed claim against Michigan State University, covering 1.03 million current and former students and staff, appeared on ransomware.live and reached news reporting about 16 hours later; it sits beside Florida's student aid website breach putting 4.9 million people at risk and Interlock's two school district leaks. These are separate actors with no evidence of coordination.

    confidence 0.50
    Wrong if
    MSU denying an intrusion, or the claimed victim count proving inconsistent.
    Watch
    An MSU statement and whether BYOD publishes data.

Nothing to seeNo link was found between the day's ransomware victims and the edge CVEs. Sageworks and Akira, Polymarket and the Cursor extension, and the supermarket refrigeration incident are co-occurrence only, with no shared entity and no causal evidence. The 41-item Mastodon cluster at 04:00 is people waking up, not a signal. The WARP PANDA ESXi zero-day claim is single-source and unconfirmed, and one day is not a baseline for any trend you are about to put in a slide.

Deep read: 343 items read

§4

Forecast

Dated, falsifiable calls. We mark our own homework in public.

Today’s calls

  1. DueFri 6 Nov72h

    NVD publication resumes in these feeds, with at least 20 new CVE entries ingested, by 6 November 2026.

    Wrong ifNo new NVD items in the feed after 1 November 21:05 by end of 6 November.

    f-2026-11-03-1

    Confidence0.80
  2. DueTue 10 Nov7d

    The Gentlemen do not post another day of 20 or more leak-site items between now and 10 November 2026; the 2 November burst was a batch release, not a new rate.

    Wrong ifAny single day to 10 November with 20 or more The Gentlemen items across ransomware.live or ransomlook.

    f-2026-11-03-2

    Confidence0.65
  3. DueFri 13 Nov10d

    At least one of Everstake's two scheduled releases (582.69 Gb due 10 November, 473.05 Gb due 12 November) appears on the leak sites, with the victim named, by 13 November 2026.

    Wrong ifBoth entries still redacted or the schedule silently slipped past 13 November with no publication.

    f-2026-11-03-3

    Confidence0.50

Scorecard

1 right · 2 wrong

  1. wrong

    At least one of N0n's anonymised 'Company #104' to '#109' entries is replaced by a named victim on ransomware.live or ransomlook.

    Called Fri 23 Oct · due Mon 2 Nov

    The due date has passed. The only N0n items are new anonymised entries (#113-#117 and #118-#122, all with placeholder names), and none shows #104-#109 being named.

  2. right

    CISA adds a third Ivanti Connect Secure CVE from the reported chain to the KEV catalogue by 8 November 2026.

    Called Sun 1 Nov · due Sun 8 Nov

    SecurityWeek reports CISA added two more Ivanti Connect Secure chain CVEs, CVE-2026-111357 and CVE-2026-118527, to KEV. This assumes the call's earlier two chain CVEs were already listed, so these are the third and fourth.

  3. wrong

    TheGentlemen posts fewer than ten new leak-site claims on ransomware.live between 2 and 8 November 2026, confirming the 31 October figure as a single burst rather than a new operating rate.

    Called Sun 1 Nov · due Sun 8 Nov

    On 2 November alone ransomlook shows about 24 new the gentlemen leaks, and the tempo report shows 50 claims in 7 days. That is already well over ten.

Glossary of terms you will be expected to nod at

Agentic (n.)
Capable of doing the boring middle part of an intrusion at three in the morning without asking for a day rate, a kick-off call, or a statement of work. Formerly: a junior. See also: Chainalysis, who had one visit.
§5

From our edge

Live attacks on our own servers in the last 24 hours. Yes, really ours.

Our own edge sensor profiled 22 hostile sources in the last 24 hours — 15 scanners, 4 bot crawlers, 2 brute forcers and 1 targeted operator — sending 1,741 requests between them, of which 4 went past reconnaissance into exploitation attempts, out of 5,262 addresses seen from 95 countries, 1,564 of them for the first time. The day's most determined visitor, 84.17.60.236 at hostility 7.7 out of 10, sent 20 requests hunting /login/wp-includes/wlwmanifest.xml and /login/xmlrpc.php and reached credential-access on its kill chain; five addresses shared a single behavioural fingerprint while asking for /.git/config, /.env and /.env.production, which is one operator with a rotation budget; 34.156.56.208 fired 123 requests for AWS and Docker credential files in about half a second; and two addresses sprayed /hnap1 at us in the shape of CVE-2015-2051, a named exploit from 2015, because the internet never throws anything away. The most confident campaign match of the day, at 45 per cent, was a WordPress plugin vulnerability scanner, which is precisely the audience that yesterday's hundred WordPress plugin CVEs were written for. Raw logs never left the server; only these findings did.

22hostile sources profiled
1,741requests from them
4reached exploitation
3shared-DNA clusters
8.2/10top hostility
5,262addresses at our edge
1,564first seen today
95countries

The WTF list

  1. WTF #1

    The most determined visitor

    84.17.60.236, a targeted operator (hostility 7.7/10), sent 20 requests to one of our sites, probing for /login/wp-includes/wlwmanifest.xml, /login/xmlrpc.php, /login/blog/wp-includes/wlwmanifest.xml, /login/web/wp-includes/wlwmanifest.xml.

    Why it’s weirdHighest hostility on the board; its kill chain reached credential-access.

    attribution: unknownTue 3 Nov, 01:00 UTC

    Detail
    kill chain: reconnaissance → weaponization → exploitation → credential-access · DNA 82a939927b85
  2. WTF #2

    Same DNA, different addresses

    35.200.118.164, a scanner (hostility 6.5/10), sent 170 requests to one of our sites, probing for /.git/config, /.env, /.env.local, /.env.production.

    Why it’s weird5 IPs share one behavioural fingerprint (DNA dfef4501b216): one tool or one operator rotating addresses: 158.23.85.214, 172.182.199.29, 20.194.27.148, 35.200.118.164, 143.244.57.84.

    attribution: unknownTue 3 Nov, 01:00 UTC

    Detail
    kill chain: reconnaissance → weaponization · DNA dfef4501b216
  3. WTF #3

    Trying CVE-2015-2051 on us

    2 addresses sent 2 requests shaped like CVE-2015-2051 exploitation (/hnap1), including 157.230.23.187, 165.227.172.91.

    Why it’s weirdA known, named exploit still being sprayed at the internet; our edge saw it today, whatever the patch notes say about the year.

    attribution: unknown

    Detail
    probe family: dlink-hnap
  4. WTF #4

    The strangest shopping list

    20.218.73.251, a scanner (hostility 6.1/10), sent 68 requests to one of our sites, probing for /wp-content/plugins/hellopress/wp_filemanager.php, /this_is_a_new_hello_world.php, /gtc.php, /simple.php.

    Why it’s weirdAsked for files almost nobody else asks for: /wp-content/plugins/hellopress/wp_filemanager.php, /this_is_a_new_hello_world.php, /gtc.php, /simple.php, /dairg.php.

    attribution: Automated vulnerability scanner or reconnaissance tool (sensor's best match, 50%)Tue 3 Nov, 01:00 UTC

    Detail
    kill chain: reconnaissance → weaponization · DNA c67915f564de
  5. WTF #5

    In a hurry

    34.156.56.208, a scanner (hostility 5.4/10), sent 123 requests to one of our sites, probing for /.aws/credentials.bak, /.docker/.env, /.aws/config, /.aws/.env.

    Why it’s weirdFired 123 requests in about 0.5 seconds.

    attribution: unknownTue 3 Nov, 01:00 UTC

    Detail
    kill chain: reconnaissance → weaponization · DNA 18e39c4b41fb
  6. WTF #6

    Somebody we might know

    74.162.157.79, a scanner (hostility 5.8/10), sent 162 requests at 2 a minute to 2 of our sites, probing for /wp-content/plugins/hellopress/wp_filemanager.php, /this_is_a_new_hello_world.php, /1index.php, /wp-content/plugins/index.php.

    Why it’s weirdThe sensor's most confident campaign match of the day: WordPress plugin vulnerability scanner / Nuclei / WPScan at 45%.

    attribution: WordPress plugin vulnerability scanner / Nuclei / WPScan (sensor's best match, 45%)Tue 3 Nov, 01:00 UTC

    Detail
    kill chain: reconnaissance → weaponization · DNA bccc30d19899
§6

Tempo

Who is posting faster than their own normal. Computed from the Signal graph, not written by the editor.

217claims, 7 days
39of 269 groups active
thegentlemen50 this week · surging
qilin27 this week
SilentRansomGroup11 this week · surging
  • Surgingthegentlemen — 50 claims in 7 days against 25.0 expected from its own prior rate (p < 0.001)
  • SurgingSilentRansomGroup — 11 claims in 7 days against 2.5 expected from its own prior rate (p < 0.001)
  • Surginggenesis — 10 claims in 7 days against 2.3 expected from its own prior rate (p < 0.001)
  • Surgingrhysida — 7 claims in 7 days against 1.5 expected from its own prior rate (p < 0.001)
  • SurgingBooba Project — 7 claims in 7 days against 2.9 expected from its own prior rate (p = 0.028)
  • Surgingpayload — 5 claims in 7 days against 1.7 expected from its own prior rate (p = 0.030)

Gone quiet: gunra (60 days), lynx (66 days), AiLock (69 days), abyss (69 days).

Sector pressure, 30 days

  • UpTransportation — 2.1× its own prior share (72 claims, 34 expected)
  • UpEducation — 2.1× its own prior share (56 claims, 27 expected)
  • DownAgriculture and Food Production — 0.7× its own prior share (22 claims, 34 expected)

23% of claims carry no sector label.

Brand churn

Of the 231 leak-site brands first seen since 2021-09-30, 76% were still claiming 30 days after their first post and 70% were still claiming 60 days after their first post and 68% were still claiming 90 days after their first post and 62% were still claiming 120 days after their first post (Kaplan-Meier; quiet = 60 days without a claim). 18% of the 213 old enough to judge posted on one day and never again; 39% made at least half their claims in their first week. New brands this month: 11 — newest: everstake (2026-10-29, 7 claims), Shiba (2026-10-29, 5 claims), Whitearmor (2026-10-23, 19 claims).

Brands born before coverage began are excluded (their true age is unknown); a rebrand counts as one brand going quiet and another being born.

Targeting

Over the last 180 days, among brands with at least 20 labelled claims: safepay put 32% of its claims on Germany (6.2× every other brand's share); N0n put 33% of its claims on Transportation (7.3× every other brand's share); genesis put 95% of its claims on the United States (2.6× every other brand's share). Widest spread: nightspire across 10.6 effective sectors.

Lift = the brand's share of a sector or country over every other brand's share in the same window; exact binomial tails, Bonferroni-adjusted for the categories tested. Claims with no sector or an Unknown/Other label are excluded.

Posting clock

Over the last 180 days, among brands with at least 30 claims: clop released 94% of its claims on Wednesdays, half of them in 13:00–15:59 UTC; safepay released 63% of its claims on Mondays, half of them in 18:00–19:59 UTC; akira posted only 1 of 195 claims at weekends and half its claims in 12:00–13:59 UTC; half of play's claims appeared in 18:00–20:59 UTC. Leak sites overall are 32% quieter per weekend day than per weekday.

all brands by hour of day, 00–23 UTC; half of all claims land in 12:00–19:59 UTC.

Times are when trackers first saw each claim on the leak site, in UTC: not the attack time, not the operator's clock, and a site that is only reachable at certain hours shows the same pattern; batch releases show as spikes. Hour-of-day concentration: Rayleigh test; weekday shares: exact binomial against every other brand, Bonferroni-adjusted.

Leak-site claims as discovered by trackers, not confirmed breaches. Surprise compares the last 7 days with the group's own prior 173 days (Poisson). Victim claims: ransomware.live and leak-site trackers

Today’s numbers as JSON →

§7

Ride-alongs

Free, live tools. No sign-up, no sales call.

§8

Colophon

How this edition was made, plainly.

Edition 37 was made by machine and checked by code. Every item published by 11 public threat feeds in the 24 hours to 05:05 UK (343 items) was read in full: statistics computed in code, each slice read by Claude Sonnet, the structure synthesised by Claude Opus. Claude Opus then edited the paper. Every story cites the items it rests on; stories whose receipts did not resolve, or whose numbers were not in their sources, were dropped before publication (0 today). Forecasts are graded in public in later editions. 'From our edge' is live attack data from our own servers. Cost of this edition: $0.73. It is written by machine in the voice of ScottG, author of Everything Is Probably Fine, and nobody edited it by hand, which is either the point or the problem.

343items read
24hour window
$0.73cost of this edition
05:05 UTCgenerated

Sources this morning

  • NVD100ok
  • RansomLook56ok
  • Signal Atlas44ok
  • Security news43ok
  • infosec.exchange38ok
  • ransomware.live19ok
  • AlienVault OTX13ok
  • RSS10ok
  • signal:ioc-family10ok
  • atlas9ok
  • Ransomware tempo · Ninja Signal graph1ok

ITEMS READ: 343SOURCES: 11STORIES DROPPED FOR LACK OF RECEIPTS: 0COST: $0.73STATUS: NOT ESPECIALLY FINE

OWNER: the machineEDITOR: the rulesCALLS OPEN: 3NEXT REVIEW: tomorrow 06:00 UK