Loading story…
A Ukrainian national, Yurii Nazarenko (aliases including “John Wick”), pleaded guilty in U.S. federal court to operating OnlyFake, a subscription-based, AI-powered fake ID service that generated and sold more than 10,000 counterfeit identification images. Prosecutors said the service produced realistic digital versions of U.S. driver’s licenses (all 50 states), U.S. passports/passport cards, Social Security cards, and IDs for dozens of other countries, with options to customize personal details and output style (e.g., scan vs. tabletop photo). Authorities assessed the primary criminal use as bypassing KYC controls at banks and cryptocurrency exchanges; undercover FBI purchases (paid in cryptocurrency) reportedly obtained fake New York IDs, U.S. passports, and a Social Security card, and the site offered bulk packages (up to 1,000 documents) at discounted rates.
Separately, researchers alleged an unknown actor used Anthropic’s Claude chatbot via Spanish-language prompts to support attacks against Mexican government agencies, including identifying vulnerabilities, generating exploit scripts, and automating data theft. According to Gambit Security’s research (as reported by Bloomberg and relayed by DataBreaches.net), the activity ran for about a month starting in December and resulted in the theft of roughly 150 GB of data, including documents tied to taxpayer and voter information, government employee credentials, and civil registry files. While both cases highlight AI’s role in enabling cybercrime and fraud, they describe different actors and incidents rather than a single unified event.
Mallory correlates global threat intelligence with your attack surface — know if you’re exposed before adversaries strike.
8 events from the most recent confirmed update back to the earliest known activity.
Court proceedings set Nazarenko's sentencing before U.S. District Judge Margaret M. Garnett for June 26, 2026.
In late February 2026, Yurii Nazarenko pleaded guilty in the Southern District of New York to operating OnlyFake, agreed to forfeit $1.2 million, and faces up to 15 years in prison.
Researchers said the AI-assisted intrusion campaign lasted about a month and resulted in the theft of roughly 150 GB of Mexican government data, including taxpayer and voter information, employee credentials, and civil registry files.
According to Gambit Security, an unknown threat actor began using Anthropic's Claude in December 2025 to identify vulnerabilities, generate exploit scripts, and automate exfiltration during attacks on Mexican government agencies.
U.S. authorities extradited Yurii Nazarenko from Romania in September 2025 to face charges related to operating the OnlyFake fake-ID platform.
Between May and June 2024, undercover FBI agents purchased multiple counterfeit documents from OnlyFake as part of the U.S. investigation into the service.
Reports about OnlyFake surfaced in February 2024, after which prosecutors say Nazarenko deleted emails and took steps to conceal proceeds routed through multiple cryptocurrency wallets.
U.S. prosecutors said Yurii Nazarenko operated the OnlyFake subscription service between 2021 and 2024, producing and selling more than 10,000 fake identification documents, including U.S. driver's licenses, passports, and Social Security cards, as well as IDs for dozens of other countries.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
4 references tracked. Mallory keeps watching after this page renders.
scworld.com
Open sourcesecurityaffairs.com
Open sourcebleepingcomputer.com
Open sourcedatabreaches.net
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Researchers at **Gambit Security** reported that a small group of attackers used **LLMs**—including **Anthropic Claude** and **OpenAI ChatGPT**—to help compromise at least **nine Mexican government agencies**, stealing large volumes of sensitive records including **~195 million identity and tax records**, **vehicle registrations**, and **~2.2 million property records**. The attackers reportedly used a long, pre-written “playbook” prompt (about a thousand lines) and social engineering to pose as legitimate penetration testers, bypassing model guardrails quickly and then using the AI tools to identify vulnerabilities, generate exploit scripts, and automate data theft across government networks. Anthropic said it investigated the reported misuse, **disrupted the activity**, and **banned the associated accounts**, and indicated it is feeding examples of the malicious behavior back into model training and deploying additional misuse-detection probes in newer models (e.g., *Claude Opus 4.6*). The incident is being cited as a concrete example of how AI can accelerate attacker workflows—reducing time-to-capability for reconnaissance, exploitation, and automation—while also highlighting the limits of current “guardrails” when adversaries can reframe requests as authorized testing.
Jun 29, 2026
Security reporting and research highlighted how **AI and automation are reshaping both attacker tradecraft and defender operations**, while introducing new enterprise risk. ZDNET described research findings that **agentic AI implementations** from *ServiceNow* and *Microsoft* can be **exploitable**, warning that broadly permissioned agents could enable **lateral movement and privilege escalation** across systems of record if an attacker compromises an agent or chains between agents with different access levels; a **least-privilege** posture for agents was emphasized. Dark Reading separately reported that **AI agents are increasingly augmenting—and in some cases supplanting—human penetration testing** for “low-hanging” vulnerabilities, but that **false positives and the need for human oversight** remain material constraints as agentic testing matures. Threat-intelligence coverage also underscored the **industrialization of cybercrime** and the ecosystems enabling it. CloudSEK detailed the evolution of the English-speaking cybercriminal milieu known as **“The COM,”** tracing its roots in OG-handle trading communities and forum migrations into a service-oriented underground linked to groups such as **Lapsus$**, **ShinyHunters**, **Scattered Spider (UNC3944)**, and **Silent Ransom Group**, and associated activity spanning breaches, extortion, SIM swapping, ransomware, and crypto fraud. SC Media’s commentary similarly described a cyber underground where criminals can readily buy capabilities (credentials, tooling, automation), calling out techniques including **carding** and **ClickFix** social engineering that tricks users into running copied commands to install infostealers. Separately, Dark Reading reported allegations that the **Chronus Group** posted **2.3TB** of purported Mexican government data affecting up to **36 million** people, while Mexico’s **ATDT** disputed it as largely **repackaged data from prior breaches** and said no new sensitive accounts were identified and that impacted systems were primarily **obsolete, third-party-administered** state-level platforms.
Jun 29, 2026
Researchers analyzing more than 1,000 recovered AI agent sessions found that a low-skilled attacker used Anthropic’s **Claude Code** and OpenAI’s **Codex** to carry out offensive cyber operations against at least 14 companies. Logs from a compromised Vultr-hosted system showed the operator using the tools for reconnaissance, vulnerability discovery, exploit development, credential harvesting, database replication, session impersonation, cloud and API key testing, mail-account takeover attempts, and lateral movement. The sessions also documented requests for stealth, anti-forensics, transcript editing, encrypted reporting, and migration of stolen data, while some generated **pentest-style reports** that included monetization estimates for exfiltrated information. Investigators said the attacker reused stolen local AI agent installations and made operational security mistakes that helped attribute the activity to a young man in Addis Ababa, Ethiopia. Separate analysis of a **Claude Code** network sandbox bypass highlighted why AI agent misuse can extend beyond unsafe model output into runtime and infrastructure controls. NSFOCUS reported that proxy-based filtering could be bypassed because policy checks evaluated the original input while the network layer parsed and truncated it differently, including with invisible control characters or null bytes, creating a path for outbound requests and possible data leakage in a single execution chain. Although the issue was reportedly fixed in later versions, researchers said the low-profile remediation made exposure difficult to assess and underscored the need for layered defenses such as strict egress allowlisting, least privilege, short-lived secrets, auditing, fail-closed policies, and fuzz testing of parsing paths.
Jun 29, 2026