Threat Intelligence
Adversary methods, active campaigns and evidence-based threat research. No fear. Just analysis.

Unit 42 Details Covert HTTP/3 Espionage Tools PerfectPioneer and KUKURAT Targeting Governments
Palo Alto Networks' Unit 42 describes a multi-stage campaign against Southeast Asian government networks using previously undocumented remote-access tools and HTTP/3 communications.

Fugaku Supercomputer Shut Down After Intrusion; Up to 4,597 Research Users Potentially Affected
Japan's RIKEN says attackers used a public status dashboard to enter Fugaku's internal operations network; account details and external access keys may have been…

AI Testing Spillover: Trajectory Labs and Epoch AI Document Agents Probing a Real Website
Newly public incident reports describe AI agents probing Epoch AI's public website during third-party evaluations; the evaluator accepts responsibility for inadequate consent and isolation.

Google Threat Intelligence Tracks Multi-Agent Attack Frameworks and Stolen AI Service Access
GTIG and Mandiant's November 16 research describes attacker use of agentic frameworks, AI credential abuse and automation; it does not claim every intrusion is…

California Sues OpenAI Over Alleged AI-Agent Cybersecurity Failures; Company Disputes Singling It Out
California's attorney general alleges inadequate safeguards during AI-agent testing led to cyber incidents. OpenAI says it has expanded safeguards and continues working with regulators.

Bitrise Restores Most Services After Cloud Security Incident; Credential Rotation Still Recommended
Bitrise says most services are operating again after November 13 unauthorized access to an internal environment. Customers are still advised to rotate linked secrets…

Optimizely Script Compromise Spreads Malicious Code Across 178 Organizations’ Websites
Updated November 18: Censys maps infrastructure linked to the Optimizely script compromise, separating observed web properties from confirmed infections.
Barracuda Details 44-Hour Cloud Intrusion Traced to Unrevoked Third-Party AI Vendor Key
Barracuda says a cloud access key left active after a third-party AI project was compromised, enabling a 44-hour intrusion before detection and response.
ScreenConnect SCKit Attacks Hit Hundreds of Endpoints; ConnectWise Urges Extension Shutdowns
Updated November 17: ConnectWise has completed its investigation, but affected extensions must remain disabled and technician clients still require updates.
Kava Reports 53.9 Million KAVA Taken in HARD Lend Exploit; Lending Shutdown Vote Planned
Kava's November 16 incident report says a collateral-accounting flaw enabled the withdrawal of 53.9 million KAVA from HARD Lend; the team plans a governance…