Resources

Threat intelligence, research, and everything you need to understand preemptive security.

All resources

Press Enter to search or Esc to close

Attack Observatory

What attackers are trying right now

Real, unsolicited login attempts against monitored infrastructure, captured as they happened. Someone scanned, picked a device, and tried to authenticate. This is what they were after — and exactly what they typed.

Loading live capture data…

The last 30 days at a glance

The same capture, summarised: which devices drew the most interest, which credentials were tried most often, and where the traffic came from.

Over the last 30 days, PacketViper recorded 14,588 unsolicited login and reconnaissance attempts from 1,801 unique source IP addresses in 85 countries, against 21 distinct device types. 361 different username and password combinations were tried. Last updated 2026-07-27 19:50 UTC.

The most frequently targeted devices were Cisco Catalyst 2960-X, Microsoft Active Directory (WS2022), VMware ESXi 7.0, Microsoft Exchange Server 2019 and Dell iDRAC9.

Most targeted devices

DeviceAttemptsUnique sources
Cisco Catalyst 2960-X5,718582
Microsoft Active Directory (WS2022)3,07288
VMware ESXi 7.01,01053
Microsoft Exchange Server 2019876169
Dell iDRAC9702203
Fortinet FortiGate563109
Cisco ASA 5516-X446119
Grafana OSS36564
F5 BIG-IP LTM330131
Juniper SRX30024371
Kubernetes 1.28 API22432
Check Point R81 Gateway21491

Most-attempted credentials

UsernamePasswordAttempts
adminadmin32
useruser22
root(none)20
rootroot19
ftpftp17
*1$416
ubntubnt16
rootpassword16
defaultdefault16
;�admin.$cmd����hello�?>�admin.$cmd����isMaster�…(none)15
rootzlxx14
rootvizxv14
root123413
adminadmin123413
rootxc351113

Where the attempts came from

CountryAttemptsUnique sources
United States6,194764
Iran2,1457
Pakistan1,35551
Russia82966
Ukraine60381
Britain (UK)576218
China53393
Netherlands23632
Korea (South)20116
Germany16527
Romania16517
United Arab Emirates15048

Networks the attempts originated from

OperatorASNAttempts
Iran Telecommunication Company PJSAS582242,126
GOOGLE-CLOUD-PLATFORMAS3969821,861
MICROSOFT-CORP-MSN-AS-BLOCKAS80751,027
Akamai Connected CloudAS63949864
Cyber Internet Services Pvt Ltd.AS9541533
Moon DcAS208485469
RostelecomAS12683430
Enjoyvc Cloud Group Limited.AS136897412
DIGITALOCEAN-ASNAS14061398
AMAZON-02AS16509292
CHINA UNICOM China169 BackboneAS4837254
AMAZON-AESAS14618220

How this data is captured

Preemptive defense turns an attacker’s first move into intelligence — before anything of value is touched.

Moving, not static

Static defenses get mapped, and a map is all an attacker needs. Automated Moving Target Defense keeps changing what can be seen and reached, so the picture an attacker assembles is stale before they can act on it. Effort gets spent on ground that has already shifted.

The attempt is the signal

A blocked connection is not a solved problem — it is the beginning of an investigation. Each attempt here carries context: which device drew interest, from where, on which service, and with which credentials. That is what a block count can never tell you.

Credentials, captured intact

The usernames and passwords shown are recorded exactly as submitted. They are the live contents of real attack dictionaries — and a direct answer to the question every operator should ask: would these have worked on my network?

How to read this page

Method

Window: a rolling 30 days of captured activity.
Refresh: the page rebuilds hourly from the live capture store.
Attack Target: the device the source was attempting to authenticate against.
Attempts From Source: that source’s full total for the window, not just the rows shown.

Limits worth stating

Country is derived from the regional internet registry allocation for the source network. That is a registration fact, not a physical location — traffic routed through hosting providers or VPNs will show the registered country of the network, not the operator.
A source address is not a culprit. Many are compromised third-party hosts being used by someone else.
Volume is capped for readability at the 25 most recent attempts per source, so one high-volume scanner cannot fill the table.

Common questions

What usernames and passwords do attackers try most often?

Across the last 30 days the most-attempted combinations were: admin / admin (32 attempts); user / user (22 attempts); root / no password (20 attempts); root / root (19 attempts); ftp / ftp (17 attempts). These are the live contents of real attack dictionaries, recorded exactly as submitted.

Which devices do attackers target most?

The most frequently targeted devices were Cisco Catalyst 2960-X, Microsoft Active Directory (WS2022), VMware ESXi 7.0, Microsoft Exchange Server 2019 and Dell iDRAC9. In total 21 distinct device types drew unsolicited authentication or reconnaissance attempts.

Which countries do these attacks come from?

By volume the largest sources were United States, Iran, Pakistan, Russia and Ukraine. Country is derived from the regional internet registry allocation for the source network, which is a registration fact rather than a physical location: traffic routed through hosting providers or VPNs shows the registered country of that network, not the operator behind it.

Is this real attack data?

Yes. Every row is an unsolicited attempt against monitored infrastructure, recorded as it happened. Nothing is simulated or synthesised. The page refreshes hourly and shows a rolling 30-day window.

Does a source IP address identify the attacker?

No. Many source addresses are compromised third-party hosts, or shared hosting and VPN exit points being used by someone else. A source address indicates where traffic arrived from, not who is responsible for it.

See It On Your Network

This is one network’s view. Now picture your plant floor.

PacketViper runs inline in OT and IT networks — agentless, non-intrusive, and surgical enough to scope enforcement to a single device instead of an entire subnet.