Radar
Machine-filtered, unreviewed, newest first.How this works.
Last updated 2026-10-16 00:33 UTC · 58 items · filter gate-2026-10-12.3
What the markers mean
- ● active
- Someone is being attacked with this. Check your own environment.
- ● exploitable
- A real flaw with an identifier or working exploit. No reported use yet.
- ● technique
- A new attack or defence method, demonstrated by researchers.
- ○ context
- Analysis, guidance or commentary. Background reading.
CVE badges carry real data: severity and publication date from NIST NVD, and KEV means CISA has stated the flaw is known to be exploited, with a federal remediation deadline. A KEV marking outranks everything else here — it is the one authoritative signal on this page.
The tier markers are different: assigned by pattern matching plus a model, not by a person. They are not evidence grades — those appear only on ramblings, where a named human has traced a claim to its primary source. Treat these as triage, not as findings.
- Securing AI○ contextOrca SecurityFrontier-Level Red-Teaming From a Self-Hostable Open-Weight Model: Testing NVIDIA Nemotron 3.5 SuperOriginal research on AI attack surfaces; AI agents as attack vectors in security context
- AI in cybersecurity○ contextSnykOriginal research on AI-driven dependency fixing; AI used defensively to secure software supply chains
- AI in cybersecurity○ contextHuntressStrategic analysis; AI-powered attacks are central to the security story
- AI in cybersecurity● exploitableEmbrace The RedCVE-2026-1035728.8HIGH3 days oldVulnerability disclosure; AI (Copilot) is attack vector for RCE
- Securing AI● techniqueChainguardVulnerability disclosure; AI vulnerability in AI infrastructure
- Securing AI○ contextJFrog SecurityOriginal technical analysis on AI governance in SDLC; AI agents are central to securing development processes
- AI in cybersecurity● techniqueOWASP GenAI SecurityThreat intelligence report; AI agents causing harm via unintended internet access
- AI in cybersecurity○ contextCisco TalosOriginal research on AI-analysis evasion techniques; AI is central to both attack (malware evading analysis) and defense (AI analysis tools)
- AI in cybersecurity○ contextMicrosoft Security BlogOriginal security research on AI-native vulnerability discovery; AI used in security research (autonomous vulnerability finding)
- AI in cybersecurity● exploitableWiz ResearchOriginal security research; AI tool (Red Agent) used in defensive security operations
- AI in cybersecurity○ contextSnykOriginal security research on AI model vulnerability discovery; AI models used to find and chain vulnerabilities into breaches
- AI in cybersecurity○ contextSimon WillisonOriginal security research on rogue AI agents; AI used offensively in unauthorized bot activities
- Securing AI● techniqueRapid7Original research on AI agent security; AI central to securing agent-to-agent communication challenges
- Securing AI● techniqueCrowdStrikeTechnical analysis of evasion techniques; LLM safety classifiers are core to securing AI systems
- AI in cybersecurity○ contextSchneier on SecurityOriginal research on AI breaking historic cipher; AI used to break cryptographic system
- AI in cybersecurity○ contextHuntressThreat ranking report; AiTM is AI-driven cyber threat
- Securing AI○ contextSnykOriginal research on AI-generated security flaws; AI coding agents causing broken access control in security systems
- Securing AI● exploitableEmbrace The RedCVE-2026-656699.6CRITICAL38 days oldVulnerability disclosure; AI tool (SQL Copilot) is the system under attack
- Securing AI○ contextSnykTechnical explainer; AI agents used in application security defense
- AI in cybersecurity● exploitableWiz ResearchOriginal investigation report; AI agent used to investigate cyber attacks
- AI in cybersecurity● exploitableSentinelOneOriginal research on evaluating agentic systems; AI central to security analysis of SOC tools
- AI in cybersecurity● exploitableHuntressOriginal security research; AI used in attack vector (ChatGPT GPTs delivering malware)
- Securing AI○ contextGitHub Security LabOriginal research on AI-driven vulnerability discovery; AI used as security tool to find Android vulnerabilities
- AI in cybersecurity● techniqueChainguardVulnerability disclosure with original AI findings; AI used in vulnerability discovery
- Securing AI● exploitableWiz ResearchAI environments targeted by credential theft, breach material to security story
- AI in cybersecurity● techniqueGitHub Security LabOriginal technical findings on AI-driven fuzzing pipeline; AI used in defensive security process for vulnerability discovery
- AI in cybersecurity○ contextCisco TalosOriginal security research on AI-driven malware; AI enables autonomous C2 in offensive operations
- AI in cybersecurity● techniqueCisco TalosOriginal research on AI malware detection methodology; AI artifacts in attack vectors and defense mechanisms
- AI in cybersecurity○ contextTrail of BitsOriginal security research on AI-assisted audit tools; AI used to find zero-knowledge VM vulnerabilities
- Securing AI● techniqueUnit 42Original security research on AI agent vulnerabilities; Prompt injection in AI infrastructure security
- Securing AI○ contextWiz ResearchOriginal research on AI detection engine; AI agents' malicious intent in security context
- AI in cybersecurity○ contextCisco TalosOriginal security research on ransomware groups; AI used offensively by Qilin in cyber operations
- AI in cybersecurity○ contextTrail of BitsOriginal research on AI patching flaws; Critiques AI's role in vulnerability remediation
- Securing AI● techniqueSentinelOneOriginal research on AI safety bypass; AI security of models under attack
- AI in cybersecurity○ contextUnit 42Original technical research on behavioral clustering model; AI-driven threat detection method in cybersecurity
- AI in cybersecurity○ contextESET WeLiveSecurityOriginal research on evading AI analysis; AI used offensively to bypass security tools
- Securing AI○ contextSnykTechnical explainer; AI in securing code generation controls
- Securing AI● exploitableWiz ResearchOriginal security research on AI infrastructure vulnerabilities; SECURING_AI - cloud AI infrastructure exposed through authentication bypass and IAM theft
- AI in cybersecurity○ contextTrail of BitsVulnerability disclosure in Lean; AI used in code review process to discover bug
- AI in cybersecurity● exploitableUnit 42Threat intelligence report; AI used in offensive operations (data exfiltration)
- AI in cybersecurity○ contextHuntressOriginal security research on AI-driven phishing kit; AI (AiTM) is central to the attack method
- Securing AI○ contextJFrog SecurityTechnical analysis of compliance risks from AI agents; AI agents directly impact audit trails and regulatory compliance
- Securing AI○ contextJFrog SecurityTechnical explainer on AI governance; AI central to security story about governance in DevOps pipelines
- AI in cybersecurity● exploitableUnit 42Original incident report on AI-driven attack; AI used offensively in breach
- AI in cybersecurity○ contextTrail of BitsOriginal research on agent sandbox escape; AI used offensively to escape VMs
- AI in cybersecurity○ contextTrail of BitsOriginal security research on AI-assisted bug hunting; AI (Codex) used to find vulnerabilities in open-source software
- AI in cybersecurity● techniqueTrail of BitsOriginal research on AI-driven vulnerability discovery; AI used to find security bugs in zlib
- AI in cybersecurity● techniqueTrail of BitsOriginal research on AI-driven vulnerability discovery; AI used to find security flaws in open-source projects
- Securing AI● techniqueTrail of BitsAI skills are central to the security story; malicious skills attack agentic systems, scanners defend them; Original research on scanner evasion techniques
- Securing AI● techniqueGoogle Security BlogAI is central to the security story about prompt injection vulnerabilities in AI systems
- Securing AI● techniqueGoogle Security BlogTechnical explainer; AI system security against prompt injection
- AI in cybersecurity○ contextOpenSSF AI SecurityOriginal findings on CRS frameworks; AI systems finding and patching bugs · low-confidence triage
- Securing AI● techniqueGoogle Security BlogTechnical explainer on agentic security; Securing AI systems against prompt injection
- Securing AI● techniqueGoogle Security BlogOriginal security research on prompt injection mitigation; AI is the target of attack vectors
Why an item matched is the classifier's own stated reason, shown verbatim so you can judge the filter rather than trust it. Links go to the original source; we reproduce no article text.