🚨 FortiGate edge intrusions are enabling deep network compromise.
Attackers exploit vulnerabilities and weak credentials to extract config files, obtain AD service credentials, deploy RMM tools, and exfiltrate NTDS.dit hashes.
🚨 FortiGate edge intrusions are enabling deep network compromise.
Attackers exploit vulnerabilities and weak credentials to extract config files, obtain AD service credentials, deploy RMM tools, and exfiltrate NTDS.dit hashes.
How AI Assistants are Moving the Security Goalposts
https://krebsonsecurity.com/2026/03/how-ai-assistants-are-moving-the-security-goalposts/
#JamiesonO'Reilly #ALittleSunshine #LatestWarnings #TheComingStorm #SauravHiremath #RiskyBusiness #SimonWillison #OrcaSecurity #WebFraud2.0 #AIassistant #JamesWilson #MattSchlict #LauraEllis #agenticAI #AmazonAWS #Anthropic #FortiGate #microsoft #RoiNisimi #SummerYue #AIagents #ClawdBot #grith.ai #Moltbook #OpenClaw #CJMoses #Copilot #Moltbot #Claude
CyberStrikeAI used in 600+ FortiGate intrusions across 55 countries.
AI-driven exploit orchestration + 100+ integrated tools.
21 IPs tied to China-based infra.
Patch FortiGate immediately.
Full report:
https://www.technadu.com/cyberstrikeai-deployed-in-over-600-fortigate-attacks-targeting-55-countries/622100/
Bleeping Computer: Amazon: AI-assisted hacker breached 600 Fortinet firewalls in 5 weeks. “Amazon is warning that a Russian-speaking hacker used multiple generative AI services as part of a campaign that breached more than 600 FortiGate firewalls across 55 countries in five weeks. A new report by CJ Moses, CISO of Amazon Integrated Security, says that the hacking campaign occurred between […]
https://rbfirehose.com/2026/02/28/amazon-ai-assisted-hacker-breached-600-fortinet-firewalls-in-5-weeks-bleeping-computer/
Bleeping Computer: Amazon: AI-assisted hacker breached 600 Fortinet firewalls in 5 weeks. “Amazon is warning that a Russian-speaking hacker used multiple generative AI services as part of a c…
THREAT MODEL: CYBERSECURITY
for Feb. 24th, 2026
by independent journalist @violetblue
- People across the US are destroying #Flock cameras
- #Discord ’s “new” ID check tool was on a US government server
- #Arizona and #Colorado want ID scanning for all computer and phone use
- @wikipedia banned #ArchiveToday for DDoS attacks
- Some rando used #AI to hack #FortiGate
- @financialtimes unmasked a Russian oil smuggling ring (from Kremlin-controlled Rosneft) through the ring’s use of a single email server
- Another major #AotearoaNewZealand health app breach
- #TimCurry reveals Clue’s secret ending
- BBC/BAFTA leave N-word intact on tape delay
...and much more.
✨ THREAT MODEL is free to read -- please help keep it accessible to all by becoming a patron, even $1 a month makes a difference! ✨
https://www.patreon.com/posts/cybersecurity-24-151516976
#ThreatModel #ThreatModelCybersecurity #ThreatModelNewsletters #VioletBlue #infosec #cybersec #CovidIsNotOver
The categorization of the Fortinet web filter is hilarious. I got a report that a certain website is incorrectly blocked by our firewall. After looking into it, the specific website belongs to a large organization.
Different organizational levels of this large organization are categorized as:
The “correct” classification should be either general organization or simply “business”.
This is wrong on many levels. Why are advocacy groups blocked by default? Why do they have an R rating (because politics is only for adults? But then why only advocacy groups and not political organizations?).
#TIL Fortinet IPsec VPNs: they claim to support IKEv2 2FA (Certificate + EAP password), but they don't strictly enforce RFC 4739 for multiple authentication rounds.
If a native client (like strongSwan) ignores the cert and just asks for EAP, the FortiGate silently accepts it and falls back to password-only. It literally fails open!
Amazon warns Russian-speaking hacker used AI tools to breach 600+ FortiGate firewalls across 55 countries in 5 weeks via weak credentials, no exploits. 🤖
Custom AI-generated scripts parsed configs, planned lateral movement, targeted Veeam backups; shows AI now amplifies low-skill attackers at scale. 🔒
#TechNews #Amazon #Privacy #Cybersecurity #AI #Hacking #Fortinet #Surveillance #DataBreach #Infosec #Malware #Linux #FOSS #Cloud #Security #Russia #Hackers #FortiGate
Selon cyberandramen.net, un serveur mal configuré exposé début février 2026 (avec un précédent en décembre 2025) a révélé l’outillage complet d’une opération d’intrusion active ciblant des organisations sur plusieurs continents. La singularité de cette campagne réside dans l’intégration d’un pipeline LLM au cœur du workflow d’attaque pour trier les cibles, produire des plans d’attaque et maintenir plusieurs intrusions en parallèle. 🚨 Principales constatations. Un répertoire ouvert a exposé un arsenal opérant avec des victimes confirmées dans au moins 5 pays. L’opération automatise la création de portes dérobées sur des appliances Fortinet FortiGate, se connecte aux réseaux victimes, cartographie l’infrastructure interne, puis transmet les résultats à des LLM pour analyse. DeepSeek génère des plans d’attaque, tandis que Claude Code produit des évaluations de vulnérabilité et est configuré pour exécuter des outils offensifs (Impacket, Metasploit, hashcat) via un fichier de paramètres contenant des identifiants d’un grand média asiatique. Un serveur MCP inédit (« ARXON ») sert de pont vers les modèles et maintient une base de connaissance croissante par cible. Entre décembre et février, l’acteur est passé d’un outil MCP open source (HexStrike) à un système d’exploitation pleinement automatisé (ARXON + CHECKER2). Des logs indiquent que le serveur source a été utilisé pour des sessions SSH modifiant des configurations FortiGate dans plusieurs pays. Des compromis confirmés touchent une société de gaz industrielle en Asie-Pacifique, un opérateur télécom en Turquie et le média asiatique mentionné, avec des reconnaissances additionnelles visant la Corée du Sud, l’Égypte, le Vietnam et le Kenya.