HackingIncidents.ai
The ledger of AI-assisted cyberattacks · Jan 2023 – Oct 2026

AI went fromhelper to operator.

In 2024, AI ran none of the 15 attacks we can document: every one had a human at the keyboard and AI assisting. In 2026 so far, AI ran 27 of 45 on its own or nearly so. This is the public, sourced record of 105 attacks since January 2023, each graded by how strong the evidence is.

Incidents
105
Confirmed
73
Strong
29
Suspected
3

Every incident, by month: tick height is how much of the attack the AI ranhover a tick for the incident

n=105
20231 incident: too few to splitn=1
2024
n=15
2025
n=44
2026
n=45
01

Which models did attackers use?

Gemini, Claude and Claude Code lead. 43 of 105 name no model at all.

62name a model

Incidents naming each modelhatched: reported by the model's own maker

Named once each

How the model is known

The model's maker says so45
A researcher names it28
The victim says so7
Inferred from artifacts12
Not stated13

By the model's maker

n=105 incidents, 86 model mentions: one incident can name several models, so the bars add up past 62. 42 incidents were first disclosed by the model's own maker about its own platform, so these bars count disclosures, not which model attackers prefer.

02

Open-weight models or closed ones?

Closed, by far: 58 name a closed model, 7 an open one.

n=105
20231 incident: too few to splitn=1
2024
n=15
2025
n=44
2026
n=45

n=105. Open means downloadable weights (Qwen, DeepSeek, GLM, Kimi) or open-source tools; closed means a vendor's hosted model. 41 of the 53 closed-only cases were reported by the model's own maker, who can see its own platform and cannot see a model running on an attacker's machine. Years under 10 incidents show counts only.

03

Did an agent harness run the attack, and was it named?

33 of 105 ran through an agent harness. 26 name it.

One square per incidentgreen: open-source harness

Open-source harnessVendor coding agentCustom frameworkNo harnessHarness named

By kind

Named harnesses

n=105. A harness is the software that drives the model through the attack: an open-source offensive framework (PentAGI, HexStrike, Strix), a vendor's coding agent run as the operator (Claude Code), or a private framework the attackers built. 7 harness cases give the framework no name.

04

Which countries are the attackers attributed to?

China, then North Korea and Russia. 51 of 105 name no country.

China: 19China 19North Korea: 17North Korea 17Russia: 11Russia 11Iran: 4Iran 4Ukraine: 1ID: 1Pakistan: 1Serbia: 1Algeria: 1

Attributed country

Who the actor is

How the country is known

Attributed by the source41
Language or infrastructure only13
No country given51

n=105. Countries are the attribution the source gives (a government, the model's maker or the investigating firm); this site makes none of its own. Dashed circles are attributions resting only on language or infrastructure clues. One incident can name two countries.

05

What did the AI actually do?

Malware dev most often. Agentic intrusion: 7 in 2025, 23 in 2026.

H1 ’23H2 ’23H1 ’24H2 ’24H1 ’25H2 ’25H1 ’26H2 ’26to dateAll
Prepare
33122238
Get in
1429235
Run the operation
30282252
Cash out
19106

n=105, counted by half-year; a blank cell means zero incidents, and the last column runs only to 31 Oct 2026. One incident usually involves several techniques, so the columns add up past the incident count. Click a technique to filter by it, or a cell for that technique in that half-year.

06

Who got hit, and how sure is anyone?

Government and technology most. 73 of 105 are confirmed.

Sector of the targets

Evidence grade

Who disclosed it first

Read the method for what each grade requires.

n=105. Up to three sectors per incident, taken from the targets the source names.

07

Every incident, with its sources.

The ledger

105 of 105 incidents
17 Oct 2026
Vapor Panda uses ARTEX and DeepSeek against more than 200 organizations
Vapor Panda → Government and business organizations worldwide per Digg
DeepSeekARTEXChina
Confirmed
Oct 2026
BUGSTER uses AI-generated tooling against an Indian university
BUGSTER → An Indian university's online education platform per Unit 42, Palo Alto Networks
DeepSeekHermes Agent
Confirmed
Oct 2026
AiPT-67 'CyberCartel' builds CARLIY0S on PentAGI against Latin American banks
AiPT-67 (CyberCartel) → Latin American banks and financial institutions (title cites 400+) per Straiker
Kimi, Claude, GrokOpen modelPentAGI
Confirmed
Oct 2026
Suspected China-nexus operator breaches 25 servers with possible AI assistance
Unidentified suspected China-nexus operator → 25 firewalls and servers, including three systems in a US government-affiliated environment per The Injection
Claude Code, DeepSeek, KimiClosed modelClaude CodeChina
Suspected
25 Sep 2026
Chinese-speaking operator uses Cairn agents to breach company websites
Unnamed Chinese-speaking operator → E-commerce and corporate websites, including Deere, SharkNinja and Cummins-associated sites per Tech Wire Asia
DeepSeek, Claude CodeClosed modelCairnChina
Confirmed
15 Sep 2026
Attacker hijacks developer's live AI coding session to spread Shai-Hulud worm
Unnamed → A SaaS provider and its developers (about 100 internal repos) per Mandiant / Google Cloud
Model not named
Confirmed
10 Sep 2026
Chinese operators ran an AI exploit foundry and agent swarms
GTG-10007 → About fifty organizations including Southeast Asian government agencies per Anthropic
ClaudeClosed modelChina
Confirmed
10 Sep 2026
Midnight Blizzard-linked GTG-20006 automated espionage and malware rebuilds
GTG-20006 (consistent with Midnight Blizzard) → Ukrainian and European governments, defense, drone suppliers per Anthropic
Claude Code, ClaudeClosed modelClaude CodeRussia
Confirmed
10 Sep 2026
Iran-nexus GTG-30004 built OSINT profiling and obfuscated NanoDump
GTG-30004 → Israeli officials and Jewish diaspora organizations per Anthropic
ClaudeClosed modelIran
Confirmed
10 Sep 2026
ShinyHunters affiliates used Claude for mass smash-and-grab breaches
GTG-50014 (suspected ShinyHunters affiliates) → SaaS vendors, an airline, a technology provider and others per Anthropic
ClaudeClosed model
Confirmed
10 Sep 2026
GTG-50020 injected prompts into an AI vendor sandbox to steal API keys
GTG-50020 → Hotel booking and fintech platforms; about thirty AI companies per Anthropic
ClaudeClosed modelRussia
Confirmed
10 Sep 2026
Fraudulent cheap-Claude reseller harvests Anthropic credentials
GTG-50021 (alias kl1zy) → Customers seeking discounted Claude access per Anthropic
Model not namedRussia
Confirmed
8 Sep 2026
BASIN CASTLE used LLMs for recon, lures and obfuscated malware
BASIN CASTLE (BASIN, TEMP.Hex) → High-value targets per Google Threat Intelligence Group
GeminiClosed modelChina
Confirmed
8 Sep 2026
MIDNIGHT NEPTUNE used LLMs for RATs, lures and lateral movement
MIDNIGHT NEPTUNE (formerly UNC1069) → Cryptocurrency organizations per Google Threat Intelligence Group
Gemini, DeepSeekOpen modelNorth Korea
Confirmed
8 Sep 2026
PRC group chained Claude, Gemini and Codex into an exploitation pipeline
PRC-nexus espionage group (unnamed) → Government entities per Google Threat Intelligence Group
Claude, Gemini, CodexClosed modelChina
Strong
8 Sep 2026
Exposed 'Recon' framework used AI agents to harvest 23,800 secrets
Unattributed → Server-side infrastructure; cloud and AI service keys per Google Threat Intelligence Group
Model not named
Strong
8 Sep 2026
SANDWORM RELIC integrated Gemini into operations against Ukraine
SANDWORM RELIC (APT44) → Ukraine per Google Threat Intelligence Group
GeminiClosed modelRussia
Confirmed
8 Sep 2026
ShinyHunters cluster UNC6240 used AI across SaaS extortion attacks
UNC6240 (ShinyHunters) → SaaS customers per Google Threat Intelligence Group
GeminiClosed model
Confirmed
Sep 2026
Operator uses ARTEX and multiple models to breach South Korean financial institutions
Unnamed Chinese-speaking, financially motivated operator → South Korean financial institutions, including Shinhan Bank-related loan-broker infrastructure per Flare
DeepSeek, GLM-5.3, Grok 4.6Open modelARTEXChina
Confirmed
Sep 2026
Ransomware affiliates use Phantom and OpenCode in human-approved intrusions
Affiliates associated with INC Ransomware and The Gentlemen → Enterprise organizations recorded in the affiliates' exposed operational infrastructure per Alignment Incident Database
Claude, Kimi K3Closed modelPhantom
Confirmed
Sep 2026
AI agents execute full ransomware intrusion in under ten hours
Unidentified extortionist → An enterprise network breached through an exposed API endpoint, then cloud, CI/CD and secrets manager per Resultsense (via The Register, reporting Unit 42)
Model not named
Confirmed
31 Jul 2026
Eleven nations warn DPRK IT workers use real-time deepfakes in interviews
North Korean IT workers → Employers hiring remote staff per TechTimes
Model not namedNorth Korea
Confirmed
Jul 2026
CyberXero initial access broker runs 51 Claude Code agents plus PentAGI
CyberXero → Ukrainian energy and utilities, plus mass WordPress/e-commerce scanning per SOCRadar
Claude CodeClosed modelClaude CodeRussia
Confirmed
Jul 2026
Open-source AI agents breach online retailers and plant card skimmers
Unnamed financially motivated operator → Online retailers (at least 27 compromised; a US airline and hospitality firm among those named in coverage) per Hackread (reporting Gambit Security)
Claude, DeepSeek, GLMOpen modelStrix
Confirmed
Jul 2026
'Skynet' agent hijacks exposed LLM endpoints to power autonomous attacks
Unnamed operators → Exposed LLM endpoints (honeypot) and internet-facing web frameworks per Symantec (security.com)
Model not namedSkynet
Confirmed
12 Jun 2026
Attacker uses someone else's exposed Ollama server as the brain of an offensive pipeline
Unnamed (VAPT framework) → Victim's exposed Ollama server; probe targets were practice and lab ranges per Sysdig
Model not named
Confirmed
3 Jun 2026
Vibe-coded PowerShell script maps Active Directory during intrusion
Unidentified → Victim organization accessed via compromised remote credentials over RDP per Huntress
Model not named
Strong
Jun 2026
JadePuffer: LLM agent runs an entire extortion operation from Langflow exploit to data destruction
JadePuffer (Sysdig name) → Internet-facing Langflow instance, then an exposed MySQL and Alibaba Nacos production server per The Register
Model not namedJadePuffer LLM agent
Confirmed
11 May 2026
APT27 uses Gemini to build relay-network management tool
APT27 → Own operational relay box (ORB) infrastructure per Google Threat Intelligence Group
GeminiClosed modelChina
Confirmed
11 May 2026
DPRK and PRC clusters use Gemini and agent tools for scaled vulnerability research
APT45; UNC2814 → Embedded devices (TP-Link firmware, OFTP implementations) and CVE/PoC validation per Google Threat Intelligence Group
Gemini, Claude CodeClosed modelOpenClawNorth Korea
Confirmed
11 May 2026
Russia-nexus malware with LLM-generated decoy code targets Ukraine
Suspected Russia-nexus actors → Ukrainian organizations per Google Threat Intelligence Group
Model not namedRussia
Strong
11 May 2026
PROMPTSPY Android backdoor drives the phone UI through Gemini
Unnamed → Android users per Google Threat Intelligence Group
GeminiClosed modelGeminiAutomationAgent
Confirmed
May 2026
Chinese-speaking operator uses Claude harnesses against government targets
Unnamed Chinese-speaking operator → Brazilian and Colombian government entities; businesses in Brazil and the United States per CrowdStrike
Claude Code, Claude, DeepSeekClosed modelClaude CodeChina
Confirmed
May 2026
Claude Code supervisor and executor agents breach Argentine healthcare organizations
Unnamed Spanish- and English-speaking operator → At least 15 Argentine healthcare entities per CrowdStrike
Claude Code, Claude, Claude Haiku 4.5Open modelClaude Code
Confirmed
May 2026
Criminals build 2FA-bypass zero-day with AI for planned mass exploitation
Unnamed prominent cybercrime actors → Popular open-source web-based system administration tool (unnamed) per Google Threat Intelligence Group
Model not named
Strong
May 2026
Indonesian-speaking operator uses Claude Code skills to compromise government and business targets
Unnamed Indonesian-speaking operator → Indonesian government domains, another country's government domain, and private-sector entities per CrowdStrike
Claude CodeClosed modelClaude CodeID
Confirmed
May 2026
Russian-speaking operators use Claude Code to steal CRM client data
Unnamed Russian-speaking group → A US CRM vendor, SaaS and data-center businesses, and a government domain per CrowdStrike
Claude CodeClosed modelClaude CodeRussia
Confirmed
Apr 2026
China-based studio ran 20+ dating apps with AI personas to defraud users
GTG-15001 → Dating app users per Anthropic
ClaudeClosed modelChina
Confirmed
Apr 2026
A hacktivist used Claude to breach European political entities
GTG-50029 → European political parties, media, think tanks and their SaaS providers per Anthropic
ClaudeClosed model
Confirmed
Mar 2026
APT36 mass-produces 'vibeware' implants with AI tools
APT36 / Transparent Tribe → Indian government networks and related entities per Dark Reading (reporting Bitdefender 'APT36: A Nightmare of Vibeware')
Model not namedPakistan
Strong
23 Feb 2026
Fideuram chairman tricked by cloned voice and fake CEO messages into EUR 95M transfers
unknown → Fideuram - Intesa Sanpaolo Private Banking per Gulf News
Model not named
Strong
12 Feb 2026
APT31 drove Gemini with Hexstrike MCP tooling against US targets
APT31 → US-based targets per Google Threat Intelligence Group
GeminiClosed modelHexStrike-AI MCPChina
Confirmed
Feb 2026
Slopoly: likely AI-generated PowerShell backdoor in Hive0163 ransomware attack
Hive0163 → Enterprise victim of ClickFix-initiated ransomware intrusion per IBM X-Force
Model not named
Strong
Feb 2026
UAT-10147 uses PentestGPT, DeepAudit and AI-written scripts against web servers
UAT-10147 → About 170,000 target URLs; Windows and Linux web servers worldwide (Brazil, Bolivia, China, Canada, Vietnam seen) per Cisco Talos
Model not namedPentestGPTChina
Confirmed
Jan 2026
UNC1069 hits crypto fintech with fake Zoom and apparent deepfake CEO video
UNC1069 (CryptoCore / MASAN) → Cryptocurrency fintech and Web3 firms, macOS host per Google Cloud / Mandiant
GeminiClosed modelNorth Korea
Strong
Dec 2025
Hacker uses Claude Code and GPT-4.1 to breach nine Mexican government bodies
Unnamed single operator → Nine Mexican federal, state and municipal organizations incl. tax authority SAT and electoral institute INE per Gambit Security
Claude Code, ChatGPT / GPTClosed modelClaude Code
Confirmed
28 Nov 2025
AI-assisted AWS intrusion reaches admin in under 10 minutes
Unnamed actor (Serbian-language comments in injected code) → One AWS customer environment (RAG data buckets, Lambda, Bedrock) per Sysdig
Model not namedSerbia
Strong
5 Nov 2025
QUIETVAULT stealer used on-host AI CLI tools to hunt secrets
Unattributed → GitHub and NPM token holders per Google Threat Intelligence Group
Model not named
Strong
5 Nov 2025
TEMP.Zagros used Gemini to research custom malware
TEMP.Zagros (MuddyWater) → Phishing victims per Google Threat Intelligence Group
GeminiClosed modelIran
Confirmed
5 Nov 2025
North Korean UNC1069 used Gemini and deepfake lures for crypto theft
UNC1069 (MASAN) → Cryptocurrency users and industry per Google Threat Intelligence Group
GeminiClosed modelNorth Korea
Confirmed
5 Nov 2025
UNC4899 used Gemini for exploit research and tooling
UNC4899 (PUKCHONG) → Edge devices and browsers (likely) per Google Threat Intelligence Group
GeminiClosed modelNorth Korea
Confirmed
Nov 2025
COINBAIT crypto-exchange phishing kit built with AI code generation
UNC5356 overlap → Cryptocurrency exchange users per Google Threat Intelligence Group
Model not named
Strong
Nov 2025
VoidLink: advanced Linux cloud malware framework built almost entirely by AI
Unnamed single developer → Cloud workloads on AWS, GCP, Azure, Alibaba, Tencent; Docker/Kubernetes per Check Point Research
TRAE SOLOClosed modelChina
Confirmed
7 Oct 2025
Korean-language operators sought ChatGPT help for malware and C2
Korean-language operator cluster (overlaps Trellix reporting) → South Korean diplomatic missions per OpenAI
ChatGPT / GPTClosed modelNorth Korea
Strong
7 Oct 2025
Russian-speaking criminals vibe-coded RAT and stealers with ChatGPT
Russian-speaking criminal operator → Unspecified per OpenAI
ChatGPT / GPTClosed modelRussia
Confirmed
7 Oct 2025
UNK_DROPPITCH/UTA0388 operators used ChatGPT for phishing and tools
UNK_DROPPITCH / UTA0388 → Taiwan semiconductors, US academia and think tanks per OpenAI
ChatGPT / GPTClosed modelChina
Confirmed
14 Sep 2025
Shai-Hulud npm worm: malicious bash script likely LLM-generated
Unknown (Shai-Hulud) → npm maintainers and packages, wide ecosystem per Unit 42, Palo Alto Networks
Model not named
Strong
Sep 2025
Chinese state group ran largely autonomous espionage campaign via Claude Code
Chinese state-sponsored group (Anthropic: GTG-1002) → About thirty targets: tech companies, financial institutions, chemical manufacturers, government agencies per Anthropic
Claude CodeClosed modelClaude CodeChina
Confirmed
Sep 2025
Criminals discuss weaponizing HexStrike AI against Citrix NetScaler zero-days
Underground forum actors (later attributed in press to Storm-1575) → Citrix NetScaler ADC/Gateway (CVE-2025-7775, -7776, -8424) per Check Point
Model not namedHexStrike-AI MCP
Suspected
Sep 2025
HONESTCUE malware outsourced stage-two code generation to Gemini
Unattributed → Unspecified per Google Threat Intelligence Group
GeminiClosed model
Confirmed
29 Aug 2025
EvilAI: fake productivity apps built with AI-generated code
EvilAI operators → Organizations worldwide; manufacturing, government, healthcare per Trend Micro
Model not named
Strong
26 Aug 2025
s1ngularity: Nx npm compromise runs victims' AI CLIs to hunt secrets
Unknown (s1ngularity) → Developers installing malicious Nx npm versions; thousands of GitHub accounts per StepSecurity
Claude Code, Gemini CLI, Amazon QClosed modelClaude Code
Confirmed
18 Aug 2025
Credential phishing campaign with LLM-obfuscated SVG payload
Unnamed → Mostly US organizations, via compromised small-business email account per Microsoft Security Blog
Model not named
Strong
4 Aug 2025
CrowdStrike: Famous Chollima uses real-time deepfakes and GenAI at 320+ companies
Famous Chollima → Hundreds of companies hiring remote IT workers per Fortune
Model not namedNorth Korea
Strong
Aug 2025
North Korean IT workers used Claude to get and keep Western tech jobs
North Korean IT worker operatives → US Fortune 500 technology companies per Anthropic
ClaudeClosed modelNorth Korea
Confirmed
17 Jul 2025
Malicious wiper prompt shipped in Amazon Q Developer VS Code extension
lkmanka58 (alias) → Amazon Q Developer extension for VS Code (about 1M installs), release 1.84.0 per Amazon Web Services
Amazon QClosed modelAmazon Q
Strong
17 Jul 2025
Kimsuky phishing uses ChatGPT-made fake South Korean military IDs
Kimsuky (APT43) → South Korean defense-related institutions, journalists, researchers, activists per Genians Security Center
ChatGPT / GPTClosed modelNorth Korea
Confirmed
Jul 2025
Vibe-hacking data extortion of 17 organizations with Claude Code
Unnamed cybercriminal (tracked by Anthropic in its full report as GTG-2002) → At least 17 organizations: healthcare, emergency services, government, religious institutions per Anthropic
Claude CodeClosed modelClaude Code
Confirmed
Jul 2025
Koske Linux cryptominer shows signs of AI-assisted development
Unnamed (Koske) → Misconfigured JupyterLab instances per Aqua Security
Model not named
Strong
30 Jun 2025
Microsoft: Jasper Sleet uses Faceswap and voice changers for fake IT workers
Jasper Sleet (and Coral Sleet) → Western employers hiring remote IT staff per Microsoft
FaceswapOpen modelNorth Korea
Confirmed
5 Jun 2025
OpenAI bans accounts tied to North Korea-style IT worker employment scheme
Suspected DPRK-linked IT worker campaigns → Remote IT and software jobs worldwide per Korea Herald
ChatGPT / GPTClosed modelNorth Korea
Strong
Jun 2025
APT28 deployed PROMPTSTEAL, malware that queries an LLM for commands, against Ukraine
APT28 (FROZENLAKE) → Ukraine per Google Threat Intelligence Group
QwenOpen modelRussia
Confirmed
Jun 2025
BlueNoroff uses deepfaked executives in Zoom call to infect crypto employee
BlueNoroff / TA444 / Sapphire Sleet (Huntress: STARDUST CHOLLIMA) → Cryptocurrency foundation employee (macOS) per BleepingComputer
Model not namedNorth Korea
Strong
Jun 2025
PROMPTFLUX malware rewrote itself hourly through the Gemini API
Unattributed (financially motivated indicators) → Broad, industry-agnostic lures per Google Threat Intelligence Group
GeminiClosed model
Confirmed
Jun 2025
Impostor uses AI voice of Marco Rubio on Signal
unknown → Three foreign ministers, a US governor, a member of Congress per CBS News
Model not named
Confirmed
May 2025
Impersonator of Trump chief of staff Susie Wiles calls officials
unknown → Senators, governors and executives contacted as Susie Wiles per Yahoo News
Model not named
Suspected
Apr 2025
FBI warns of AI voice messages impersonating senior US officials
unknown → Current and former senior US federal and state officials and their contacts per FBI IC3
Model not named
Confirmed
Apr 2025
Unit 42: North Korean IT workers use real-time deepfakes in interviews
North Korean IT workers (Wagemole) → Employers hiring remote staff per Palo Alto Networks Unit 42
Model not namedNorth Korea
Strong
25 Mar 2025
Singapore finance director sends US$499K after deepfake Zoom call
unknown → Unnamed multinational's Singapore finance director per Mothership.sg
Model not named
Confirmed
Mar 2025
Actor used Claude to build scrapers for leaked IoT camera credentials
Unnamed sophisticated actor → Internet-facing security cameras per Anthropic
ClaudeClosed model
Confirmed
Mar 2025
Novice actor built an evasive malware builder with Claude
Unnamed novice actor → Unspecified per Anthropic
ClaudeClosed model
Confirmed
Mar 2025
Recruitment fraud campaign used Claude to polish scam messages
Unnamed actor → Job seekers in Eastern Europe per Anthropic
ClaudeClosed model
Confirmed
Feb 2025
Italian defence minister's cloned voice used to solicit tycoons
unknown → Italian business leaders (Moratti, Armani, Prada and others) per Euronews
Model not named
Strong
29 Jan 2025
APT42 used Gemini for phishing content and defense-expert recon
APT42 → Defense experts and a US defense organization per Google Threat Intelligence Group
GeminiClosed modelIran
Confirmed
29 Jan 2025
North Korean APTs used Gemini across attack phases and IT-worker fakery
Nine North Korea-backed groups → South Korea and US organizations; Western employers per Google Threat Intelligence Group
GeminiClosed modelNorth Korea
Confirmed
29 Jan 2025
PRC-backed APT groups used Gemini for recon, scripting and post-compromise research
Over 20 China-backed groups → Unspecified targets per Google Threat Intelligence Group
GeminiClosed modelChina
Confirmed
29 Jan 2025
Russian APT actors used Gemini to port and encrypt malware code
Russian government-backed APT actors → Unspecified per Google Threat Intelligence Group
GeminiClosed modelRussia
Confirmed
Jan 2025
Low-skill actor sold Claude-built ransomware for $400-$1200
Unnamed cybercriminal (tracked by Anthropic in its full report as GTG-5004) → Buyers on cybercrime forums per Anthropic
ClaudeClosed model
Confirmed
Jan 2025
Vidoc Security interviews two deepfake job candidates
unknown (possible DPRK link) → Vidoc Security Lab (hiring) per The Pragmatic Engineer
Model not named
Strong
Dec 2024
FunkSec ransomware group builds tools with AI assistance
FunkSec → Global victims across sectors, 85+ leak-site victims in Dec 2024 per Check Point Research
Model not namedAlgeria
Strong
Oct 2024
Wiz employees get deepfake voice messages from CEO
unknown → Wiz (cloud security), dozens of employees per TechCrunch
Model not named
Confirmed
19 Sep 2024
Senator Cardin targeted by deepfake of Ukraine's ex-foreign minister
unknown → US Senate Foreign Relations Chair Ben Cardin per NBC News
Model not named
Strong
15 Jul 2024
KnowBe4 hires North Korean IT worker using AI-enhanced photo
North Korean IT worker scheme → KnowBe4 (security awareness) per KnowBe4
Model not namedNorth Korea
Confirmed
Jul 2024
Ferrari executive foils CEO voice-clone call
unknown → Ferrari executive per Jalopnik
Model not named
Strong
Jun 2024
AsyncRAT campaign with GenAI-written VBScript and JavaScript
Unnamed → French-speaking users, invoice-themed phishing per HP Wolf Security
Model not named
Strong
May 2024
WPP CEO impersonated with voice clone in fake Teams meeting
unknown → WPP (advertising) per The Guardian
Model not named
Confirmed
Apr 2024
LastPass employee targeted with deepfake CEO voice call
unknown → LastPass employee per LastPass
Model not named
Confirmed
Apr 2024
TA547 delivers Rhadamanthys stealer with likely LLM-written PowerShell
TA547 → German organizations, invoice lure impersonating retailer Metro per Proofpoint
Model not named
Strong
Feb 2024
Charcoal Typhoon used ChatGPT for research, scripting and phishing content
Charcoal Typhoon → Government and defense-related targets (per source) per OpenAI
ChatGPT / GPTClosed modelChina
Confirmed
Feb 2024
Crimson Sandstorm used ChatGPT for research, scripting and phishing content
Crimson Sandstorm → Government and defense-related targets (per source) per OpenAI
ChatGPT / GPTClosed modelIran
Confirmed
Feb 2024
Emerald Sleet used ChatGPT for research, scripting and phishing content
Emerald Sleet → Government and defense-related targets (per source) per OpenAI
ChatGPT / GPTClosed modelNorth Korea
Confirmed
Feb 2024
Forest Blizzard used ChatGPT for research, scripting and phishing content
Forest Blizzard → Government and defense-related targets (per source) per OpenAI
ChatGPT / GPTClosed modelRussia
Confirmed
Feb 2024
Salmon Typhoon used ChatGPT for research, scripting and phishing content
Salmon Typhoon → Government and defense-related targets (per source) per OpenAI
ChatGPT / GPTClosed modelChina
Confirmed
Jan 2024
Arup employee tricked by deepfake video call into $25M transfers
unknown → Arup (engineering), Hong Kong office per CNN Business
Model not named
Confirmed
27 Aug 2023
Retool breached via SMS phishing and deepfaked employee voice
unknown → Retool (software), crypto customers per Retool
Model not named
Confirmed