ZeroFox Daily Intelligence Brief - September 18, 2026
|by Alpha Team
ZeroFox Daily Intelligence Brief - September 18, 2026
ZeroFox Intelligence collects, curates, and analyzes information derived from open and proprietary sources. Here is today’s daily roundup to give you and your clients an advantage over the adversary.
Brief Highlights
- Manufacturing Targeting Highlights Growing Supply Chain Risk
- OpenAI Model Circumvented API Access Restrictions During Training
- Brevo Supply-Chain Attack Injects ClickFix Malware Scripts on Customer Websites
Targeting of Manufacturing Industry Highlights Growing Supply Chain Risk
What we know: ZeroFox has observed manufacturing was the most targeted industry by ransomware groups over the past month, accounting for 167 of 883 total recorded ransomware attacks—nearly one in five—with the sector also recording the highest attack volume in the past 24 hours.
Context: Over the past month, ZeroFox recorded an average of 15 ransomware attacks targeting the manufacturing industry per day. The Gentlemen, Qilin, and Akira were the most active groups targeting the sector. Despite remaining the most targeted sector, attacks against the sector declined approximately 11 percent month-over-month, the most active groups however remained the same across both periods.
- On September 15, The Gentlemen listed 32 new victims on its leak site, with approximately 10 confirmed as manufacturing or manufacturing-adjacent organizations.
- On September 17, the Qilin ransomware group listed Thema Foundries, a Belgium-based semiconductor manufacturing company on its leak site, with a nine-day ransom deadline.
- On September 16, Akira ransomware group targeted Pilot Precision Products, a U.S.-based manufacturer of industrial broaches and precision cutting tools, threatening to upload 30 GB of corporate data if ransom demands are not met.
Analyst note: The manufacturing sector's position as the most targeted ransomware industry almost certainly reflects a deliberate strategic calculation by threat actors. This is partly driven by the sector's concentration of mid-sized suppliers embedded within the vendor ecosystems of larger enterprises, creating conditions that are very likely to produce significant supply chain disruption when compromised.They also present a high-consequence, comparatively lower-defended entry point that amplifies extortion leverage.
OpenAI Model Circumvented API Access Restrictions During Training
What happened: An internal OpenAI model attempted to bypass an API key requirement by registering a disposable email account and searching public GitHub repositories for exposed keys during training. The model obtained a working key but failed to retrieve the requested data, resulting in fabricated figures and false attribution.
Context: This incident is one of the six incidents that OpenAI published as part of the new AI model-misalignment reporting framework. In a separate incident, threat actors reportedly impersonated OpenAI with fake ChatGPT subscription-payment emails, to target users in a phishing campaign.
Analyst note: These campaigns likely show how LLM misalignment is expanding the threat landscape by creating new opportunities to exploit user trust. Moreover, as some Chinese AI companies increasingly use distillation to train their models using capabilities from OpenAI and other providers, state-linked threat actors are likely to seek less restricted models to expand their control over AI-assisted attacks.
Brevo Supply-Chain Attack Injects ClickFix Malware Scripts on Customer Websites
What we know: Attackers reportedly compromised a Cloudflare API key belonging to Brevo, a customer relationship management and digital marketing platform, and used it to inject malicious ClickFix scripts into Brevo's own websites and JavaScript files embedded on customer sites.
Context: The exposure window lasted approximately five and a half hours on September 14, 2026, potentially affecting up to 100,000 websites. Visitors were served fake Cloudflare verification pages with ClickFix instructions; on WordPress sites with a logged-in administrator, the scripts also attempted to install a persistent backdoor plugin.
Analyst note: Affected site operators will likely need to audit for rogue plugins and rotate admin credentials to close the passwordless backdoor.
DEEP AND DARK WEB INTELLIGENCE
Exploit user test_mobi: Untested threat actor "test_mobi" has advertised network access with local administrator rights allegedly associated with an unnamed U.S.-based freight and logistics services company on dark web forum Exploit. The threat actor has claimed that the company is located in Florida, employs 51–200 people, and generates USD 58 million in annual revenue.
- The threat actor did not specify the type of data accessible via the advertised access and did not provide proof of access. At the time of writing, ZeroFox has identified no public reporting or breach disclosure corresponding to a Florida-based freight and logistics company matching the details in the listing.
- In the past month, ZeroFox observed 52 ransomware attacks targeting the transportation and logistics sector, a 37 percent increase from the 38 attacks observed in the month prior to it. The Qilin ransomware group was the most active group in both periods with victims mostly concentrated in North America and Europe and Russia.
- Small and mid-sized freight companies are likely to be targeted increasingly by threat actors for initial access, as they rely on basic security controls but are connected to brokers, shippers, and larger third-party logistics providers. Access to downstream entities is likely to attract state-linked actors, particularly given current tensions around global supply chain and chokepoints.
- Additionally, access to such data is likely to be used for cargo theft and freight payment fraud. In the past, threat actors have been observed using compromised carrier systems to post fake load board listings and reroute shipments for resale.
VULNERABILITY AND EXPLOIT INTELLIGENCE
14 vulnerabilities disclosed in BIND 9: The Internet Systems Consortium (ISC) has disclosed 14 vulnerabilities affecting various versions of BIND 9 DNS software, with no reported active exploitation at the time writing. Seven of the reported flaws are rated high in severity and cover denial-of-service and resource-exhaustion conditions against the named process, while the remaining seven are rated medium in severity and include four flaws affecting DNS data integrity. ISC lists no workarounds and has patched all 14 in BIND 9.20.29 while twelve of them also affect the end-of-life 9.18 branch, for which no patch is available.
Affected products: BIND 9.11.0 through 9.18.50, 9.20.0 through 9.20.27, and 9.21.0 through 9.21.25, depending on the flaw; fixed in 9.20.29, 9.21.26, and 9.20.29-S1.
Tags: DIB, tlp:green