Russia's Sandworm Team used spear-phishing emails with malicious Office macro attachments to plant BlackEnergy3 malware inside three Ukrainian power.
Social Engineering Examples·4 sources
On December 23, 2015, Russia's Sandworm Team executed the world's first publicly confirmed cyberattack to cause a real power outage. Following a spear-phishing campaign that had begun months earlier against employees at three Ukrainian regional electricity distributors, Prykarpattyaoblenergo (Ivano-Frankivsk region), Kyivoblenergo (Kyiv region), and Chernivtsioblenergo (Chernivtsi region), attackers used stolen credentials and VPN/remote-access pathways to reach each utility's ICS/SCADA network and directly, manually operate circuit breakers at more than 50 substations within roughly 30 minutes of each other.
This cut power to approximately 225,000 customers for an estimated 3 to 6 hours. The attackers also deployed the KillDisk wiper to corrupt master boot records and destroy logs, corrupted firmware on serial-to-Ethernet field devices to block remote recovery, scheduled UPS disconnects, and ran a telephone denial-of-service attack against a utility call center to hinder outage reporting and coordination.
A joint US interagency team (DHS/NCCIC/ICS-CERT, DOE, FBI, and NERC) traveled to Ukraine and documented the event; CISA's alert was later updated in July 2021 to formally attribute the campaign to Russian state actors, and a 2020 US DOJ indictment named GRU Unit 74455 (Sandworm) officers in connection with this and a follow-on 2016 Ukraine grid attack.
Sandworm ran a spear-phishing campaign (observed as early as March 2015) against employees in the administrative and IT networks of the oblenergos, emailing malicious Word and Excel documents that appeared to come from trusted sources. Opening the document triggered a popup urging the user to "enable content"/enable macros; doing so ran an embedded VBA macro (vba_macro.exe) that dropped and installed the BlackEnergy3 malware; no software vulnerability or exploit code was used, only abuse of legitimate Office macro functionality.
Once inside, BlackEnergy3's keylogger and network-sniffer modules harvested legitimate user and administrator credentials over roughly six-plus months of reconnaissance, letting the attackers move laterally, escalate privileges, and map which VPNs and remote-access tools bridged the corporate IT network into the ICS/SCADA environment. Using those stolen credentials over VPN connections, the attackers reached the ICS network and, via existing remote administration tools (RDP, TeamViewer, rlogin) or ICS client software resembling an operator HMI, directly and manually opened breakers at more than 50 substations across the three companies within about 30 minutes of each other.
The outage was caused by this direct human interaction with SCADA/HMI systems, not by BlackEnergy3 or any malware payload itself. To delay recovery, the attackers overwrote firmware on serial-to-Ethernet communication devices at substations (blocking remote breaker recontrol), scheduled unauthorized UPS disconnects, and deployed a modified KillDisk wiper that corrupted master boot records and deleted logs on IT and, in at least one case, an HMI embedded in a remote terminal unit.
A telephone denial-of-service (TDoS) flood against a company call center simultaneously prevented customers from reporting outages and hampered internal coordination.
The lures were weaponized Microsoft Excel and Word documents emailed to individuals in the utilities' administrative and IT departments, disguised as ordinary business correspondence relevant to Ukrainian energy-sector staff (a related NCCIC/ICS-CERT technical alert describes the broader spear-phishing campaign as using MS Office attachments with "generally popular topics"; secondary vendor reporting, including Kaspersky Lab, has characterized specific BlackEnergy lure samples from the same period as mimicking a Ukrainian political-party news topic, though the primary CISA and SANS/E-ISAC sources cited here confirm the macro-delivery method without themselves naming an exact lure theme for the December 2015 attack).
The technical "tell" was the now-classic Office macro-security prompt: the document displayed as blank, garbled, or requiring a special view, with a banner urging the recipient to click "Enable Content" or "Enable Editing" to see the material. This was the standard social-engineering trick used to get a victim to manually authorize macro execution, since no exploit or vulnerability was involved.
On December 23, 2015, coordinated attackers remotely and manually operated breakers at more than 50 substations across Prykarpattyaoblenergo, Kyivoblenergo, and Chernivtsioblenergo within about 30 minutes of each other, cutting power to roughly 225,000 customers for an estimated 3 to 6 hours. Utility staff restored power substantially through manual, on-site switching after remote control was lost (operators found their own mouse cursors moving and workstations locked out).
KillDisk wiped systems and corrupted the master boot record on IT and some ICS-adjacent machines, corrupted field-device firmware, and a telephone DoS attack disabled a call center, all of which delayed recovery and forced the oblenergos to operate under constrained/manual conditions for an extended period afterward. No physical equipment damage or injuries were reported, and no ransom or extortion was involved.
The operation appears to have been a demonstration/disruption action rather than one seeking financial gain. Three additional organizations across other critical-infrastructure sectors were also found compromised in the same campaign but suffered no operational impact.
This was the watershed event proving that phishing-borne malware could be chained all the way through IT compromise into physical, real-world sabotage of critical infrastructure, not just data theft. It demonstrated that the weakest link was not a system vulnerability but an employee clicking "enable macros" on a plausible-looking Office document, and that the actual outage was caused by attackers directly using legitimate remote-access tools and stolen credentials to operate control systems as if they were authorized operators, and that no ICS-specific exploit was required.
It reshaped how the energy sector and governments think about IT/OT network segmentation, macro security, VPN/remote-access hardening, and manual-fallback procedures, and it remains the reference case cited in nearly every subsequent ICS/critical-infrastructure security standard and CISA advisory.
US CISA/ICS-CERT (joint alert IR-ALERT-H-16-056-01, updated July 20, 2021 to formally attribute the campaign to Russian nation-state cyber actors) and the SANS/E-ISAC technical report both recommend: end-user phishing awareness training, disabling Office macros by default/blocking macros from internet-sourced documents, application whitelisting on IT workstations (noted as only partially effective since it would not have stopped the ICS-side attack), network segmentation and strict access controls between corporate IT and ICS/SCADA networks, multi-factor authentication on VPN and remote-access paths, monitoring/alerting on remote administration tool use (RDP/TeamViewer/rlogin) into ICS environments, maintaining manual/offline fallback procedures for breaker operation (Ukrainian operators restored power partly via manual switching), firmware integrity checks on serial-to-Ethernet field devices, and out-of-band/backup communications for call centers to withstand telephony denial-of-service.
Social Engineering Examples. “2015 Ukraine Power Grid Attack (Sandworm/BlackEnergy)”. Accessed 19 September 2026. https://socialengineeringexamples.com/ukraine-power-grid-attack-2015
Sandworm is documented by SANS/E-ISAC's Defense Use Case as having conducted extensive, long-term reconnaissance of the three oblenergos before the attack, likely drawing on open-source information such as publicly posted ICS/RTU vendor and version lists plus routine OSINT on staff and organizational structure to identify plausible email recipients and craft convincing lures.
Publicly available ICS vendor/version listings and routine employee-level OSINT are very hard to eliminate at enterprise scale; the realistic control assumes an adversary already has this information and instead hardens what that access can reach, particularly the IT/ICS boundary addressed in Stage 6.
The group built spear-phishing payloads by embedding a VBA macro dropper (identified in independent malware analysis as vba_macro.exe) inside ordinary Microsoft Word and Excel files to install the BlackEnergy3 malware, deliberately relying on abuse of legitimate Office macro functionality rather than an exploit or vulnerability.
Macro-based droppers built from ordinary Office files are difficult to catch before delivery through static inspection alone; the practical countermeasure sits downstream, in email filtering (Stage 3) and in disabling macro execution outright (Stage 4).
Starting as early as March 2015, the weaponized documents were emailed to employees in the administrative and IT departments of the oblenergos, disguised as ordinary business or sector-relevant correspondence.
Email and attachment filtering or sandboxing for internet-sourced Office documents, combined with employee phishing-awareness training on unsolicited attachments, reduces how many weaponized documents ever reach a recipient's inbox.
Opening a document triggered a prompt urging the recipient to enable content or macros to view it; enabling macros silently installed BlackEnergy3, giving the attackers a foothold purely through the user's own action rather than any software vulnerability.
Disabling Office macros by default, or restricting execution to trusted, signed macros, and blocking macros in files that originate from the internet removes the exact 'enable content' step the attackers depended on, since CISA and SANS both confirm no exploit code was used, only macro abuse.
Over roughly six or more months, BlackEnergy3's keylogging and network-reconnaissance modules harvested legitimate user and administrator credentials, letting the attackers move laterally and map which VPNs and remote-access tools bridged the corporate IT network into the ICS/SCADA environment.
Multi-factor authentication on VPN and remote-access paths, plus monitoring for anomalous credential use and lateral movement, addresses the fact that the oblenergos' VPNs reportedly lacked two-factor authentication, and could have stopped or flagged the credential-harvesting stage before it reached the ICS network.
Using the stolen credentials over VPN, the attackers reached SCADA/HMI systems at each utility, learned each company's distinct control software, and, per SANS/E-ISAC's kill-chain analysis, developed and likely tested malicious firmware for serial-to-Ethernet field devices ahead of the operation.
Strict network segmentation between corporate IT and ICS/SCADA networks, with no shared credentials or unmonitored remote-access bridges between them, would have prevented stolen IT credentials from ever reaching SCADA/HMI systems in the first place.
On December 23, 2015, multiple operators simultaneously used existing remote-administration tools (RDP, TeamViewer, rlogin) or ICS client software to manually open breakers at more than 50 substations across the three companies within about 30 minutes of each other, cutting power to roughly 225,000 customers; this direct manual manipulation of the control systems was the objective-completion step of the attack.
Monitoring and alerting on remote-administration-tool use (RDP/TeamViewer/rlogin) into ICS environments, combined with operator procedures to detect and immediately revoke unauthorized HMI sessions, plus rehearsed manual/offline fallback procedures for breaker operation, are the last practical defenses against direct manual manipulation of control systems.
In parallel, the attackers corrupted serial-to-Ethernet firmware, scheduled unauthorized UPS disconnects, deployed a modified KillDisk wiper to corrupt master boot records and delete logs, and ran a telephone denial-of-service flood against a utility call center, all intended to delay restoration and blind both operators and customers to the scope of the outage.
Firmware integrity checks and monitoring on serial-to-Ethernet field devices, offline and immutable backups of logs and system images, and out-of-band or backup communications for call centers would have preserved recovery capability and customer communication despite the wiper, firmware corruption, and telephony flood.
Browse by what this case has in common with others in the library.
Fraudsters impersonating Ubiquiti's CEO and an outside law firm tricked its Hong Kong finance controller into wiring $46.7M abroad.
Impostors posing as Scoular's CEO and a KPMG partner convinced the controller a secret China acquisition was underway.
A spoofed email impersonating Seagate's CEO tricked an HR/payroll employee into emailing every 2015 W-2 to a scammer.
The CEO of a UK energy firm was tricked into wiring €220,000 (~$243,000) to a Hungarian account in March 2019…
A nation-state cyberweapon crossed an air gap into Iran's Natanz nuclear facility on infected USB drives carried in by trusted…
A Ukrainian-language "shared document" phishing wave used a fake reCAPTCHA "verify you are human" ClickFix page to trick victims into…
A compromised email address was used to redirect a $30,750 solar-panel-installation payment from a Wheeling.
A nation-state cyberweapon crossed an air gap into Iran's Natanz nuclear facility on infected USB drives carried in by trusted…
The CEO of a UK energy firm was tricked into wiring €220,000 (~$243,000) to a Hungarian account in March 2019…
Scammers impersonating PG&E threaten customers and small businesses with immediate service disconnection.
A widely circulated security-awareness case study (NINJIO) claims a tailgating "badge surfer" photographed passwords exposed by a clean-desk-policy.
A lone, unidentified hacker jailbroke Anthropic's Claude Code coding agent with a fake "authorized bug bounty/pentest" pretext and paired it.
Google's GTIG and Ukraine's CERT-UA documented PROMPTSTEAL (aka LAMEHUG).
Russian GRU-linked Forest Blizzard (APT28/Fancy Bear) used OpenAI's GPT-4-family LLM services to research satellite communication protocols and radar.
A Ukrainian-language "shared document" phishing wave used a fake reCAPTCHA "verify you are human" ClickFix page to trick victims into…
KnowBe4 unknowingly hired a North Korean operative for a software engineering role after he passed four video interviews using an…
A Ukrainian-language "shared document" phishing wave used a fake reCAPTCHA "verify you are human" ClickFix page to trick victims into…
Two New Jersey men ran US "laptop farms" and shell companies that let North Korean IT workers pose as American…
Operation Buckshot Yankee: a malware-laden USB drive plugged into a U.S. military laptop in 2008 spread the agent.btz worm onto…
Lazarus operators spear-phished a senior Sky Mavis engineer through a fake LinkedIn recruiting process and a spyware-laced job-offer PDF.
A suspected Chinese state-sponsored group jailbroke Anthropic's Claude Code by role-playing a "defensive security" pretext and used it to autonomously…