Malware Activity
AI-Assisted Invoice Fraud and Passkey Phishing Target Enterprise Cloud Accounts
Microsoft detailed two (2) large-scale social engineering campaigns targeting enterprise organizations through financial fraud and cloud identity compromise. In the first, threat actors sent more than one million emails between August 3 – 5, 2026, primarily targeting U.S. companies by impersonating CEOs and using AI-assisted messaging, forged email threads, fabricated ServiceNow invoices, and lookalike domains to convince accounts payable personnel to make fraudulent ACH transfers. A separate campaign active since May uses passkey-, MFA-, and SSO-themed vishing and phishing, with attackers posing as IT help desk personnel and directing employees to counterfeit Microsoft login pages or malicious device-code authentication flows. After compromising Microsoft 365 identities, attackers establish persistence by registering their own MFA methods, then use Microsoft Graph APIs and other cloud services to conduct reconnaissance, collect mailbox data, and exfiltrate files from SharePoint and OneDrive. Microsoft linked portions of the initial-access activity to Storm-3121 and Storm-3032/UNC6671, with overlaps to infrastructure and techniques associated with ShinyHunters, Falcon, BlackFile, and Helix-related extortion activity.
Threat Actor Activity
Anthropic Disrupts Illicit Distillation Campaigns Targeting Claude
Anthropic says it has disrupted seven (7) large illicit distillation campaigns against Claude run by China-based AI labs, including Alibaba, Moonshot, DeepSeek, Zhipu (Z.ai), Xiaomi, SenseTime, and MiniMax. Anthropic alleges that these labs covertly used Claude as a “teacher” to copy its capabilities into their own models by routing massive volumes of traffic through proxy services and fake accounts typically created with stolen cards, credentials, and API keys. Anthropic also reported that in some cases Moonshot and DeepSeek silently relayed their customers’ prompts to Claude, showed Claude’s answers back to users, and then saved those exchanges, which frequently contain sensitive data from individuals, multinationals, and state-linked actors, as training data. Others replayed conversations, bought transcripts from proxy resellers, or ran dedicated chain of thought (CoT) extraction pipelines, with one (1) Alibaba-linked campaign logging 151 million exchanges. To counter this, Anthropic is banning reseller and unsupported-region accounts that fail verification, encrypting and summarizing internal reasoning so stolen transcripts are less useful, and introducing “preserved thinking” controls that prevent new API accounts from editing system prompts and tools to trick Claude into revealing its reasoning.
Vulnerabilities
CISA Warns of Active Exploitation of Critical GitLab Vulnerability
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added CVE-2026-85706, a maximum-severity GitLab vulnerability, to its Known Exploited Vulnerabilities (KEV) Catalog following evidence of active exploitation. The flaw stems from missing authentication enforcement and improper path confinement in GitLab’s repository commits API, allowing unauthenticated attackers to read arbitrary files and potentially steal credentials, secrets, and other sensitive information from vulnerable servers with a single HTTP request. GitLab addressed the vulnerability in Community Edition and Enterprise Edition versions 19.3.2, 19.2.6, and 19.1 and urged customers to update immediately. Shortly after the patches were released, watchTowr observed attackers scanning the internet for vulnerable GitLab instances and advised defenders to review logs for suspicious POST requests targeting the repository commits API. CISA subsequently ordered federal agencies to remediate the vulnerability within three days under BOD 26-04 and encouraged private-sector organizations to prioritize patching due to the ongoing exploitation and significant risk posed to exposed GitLab environments.
📧 Never Miss a Briefing
Stay informed and secure. Subscribe to Ankura’s Cyber Flash Update, a bi-weekly briefing curated by our top cybersecurity experts. Receive timely insights on emerging threats, vulnerabilities and malicious actors to keep your systems secure.
Join the Cyber Flash Update community today.
© Copyright 2026. The views expressed herein are those of the author(s) and not necessarily the views of Ankura Consulting Group, LLC., its management, its subsidiaries, its affiliates, or its other professionals. Ankura is not a law firm and cannot provide legal advice.
