#detection

China-linked Fire Ant compromised Cisco IOS XR routers and TACACS servers, then rewrote the logs to stay invisible. The routers logged only "Health" heartbeats, so liveness checks passed while everything else vanished. Off-host logging catches it. https://suriq.io/blog/fire-ant-cisco-router-tacacs-credential-theft #Detection #Phishing #infosec #cybersecurity

SMOKE#SCREEN disables Windows Defender, then installs a validly signed ScreenConnect agent as its backdoor. Your signature allowlist trusts it. The tampering it does first is what gives it away. https://suriq.io/blog/smokescreen-screenconnect-rmm-defender-evasion #ThreatIntel #Detection #Phishing #Windows

Device-code phishing rose ~1,500% in 2026. It steals Microsoft 365 tokens after you sign in, so MFA and passkeys do not stop it. Hits Entra ID tenants. Fix: block the device-code flow in Conditional Access, alert on every use. https://suriq.io/blog/device-code-phishing-microsoft-365-token-theft #Detection #Phishing #infosec #cybersecurity

Malware can copy the device key behind Google Chrome passkeys and sign in as you, even with two-factor required, Unit 42 showed. The hole: sites that never check the "user verified" bit. Fix: validate that flag server-side. https://suriq.io/blog/chrome-passkey-malware-account-takeover #ThreatIntel #Detection #infosec #cybersecurity

🔴 EXPLOITED Cisco Secure Firewall Management Center ships a hardcoded password that lets a stranger log in with no credentials (CVE-2026-20316). It is being exploited now and sits on CISA's must-patch list. Apply the hotfix, then check logs for signs of entry. https://suriq.io/blog/cisco-fmc-hardcoded-password-cve-2026-20316 #CVE #Detection #CISAKEV #infosec

---------------- 🛠️ Tool =================== numbat is an endpoint visibility tool for AI agent activity, developed by perplexityai. It provides local detection, optional pre-action blocking, and forensic reconstruction of agent sessions across desktop, CLI, IDE, and gateway surfaces. 🔹 Key Features The tool observes supported agents through local hooks and plugins, OTLP/HTTP log exporters, and on-disk session artifacts. Live and at-rest activity is normalized into a single event model and evaluated by a CEL rule engine. Detection runs entirely locally. Records can be written to stdout or a local file, with optional HTTP delivery. • Live monitoring via hooks, plugins, and OTLP/HTTP exporters • Local detection with built-in CEL rules, multi-step sequence rules, and custom YAML rules • Optional blocking through supported synchronous pre-action hooks (disabled by default; only rules marked enforce: true apply) • Forensic reconstruction from on-disk session artifacts without prior numbat instrumentation • Versioned NDJSON records for events, findings, enforcement decisions, indicators, and scan summaries • Read-only artifact scanning with secret redaction; raw transcripts never included in normal output • Inventory and investigation tools for agent discovery, per-session timelines, and portable case bundles with SHA-256 manifests • Single-binary distribution for macOS, Linux, and Windows, built without cgo 🔹 Technical Implementation Installation is straightforward: download a release or use go install github.com/perplexityai/numbat/cmd/numbat@latest. Read-only inventory commands (numbat agents, numbat scan) do not install hooks or modify agent configuration. Live monitoring requires numbat hook install --agent --emit all, which starts in monitor-only mode. Hook trust requirements vary by agent and scope. For Codex user hooks, operators must review and trust the hook definition in /hooks or Settings > Hooks. Managed hooks are trusted by policy. All shipped rules are monitor-only. To enforce a detection, operators copy the shipped YAML into a controlled directory, add enforce: true, bump the version, then install with --enforce. 🔹 Use Cases • Security teams needing visibility into what AI agents execute on developer endpoints • Forensic reconstruction of past agent sessions without prior instrumentation • Compliance auditing of agent actions with versioned NDJSON records • Incident response with portable case bundles and SHA-256 manifests 🔹 Limitations Blocking is limited to supported synchronous pre-action hooks only. The coverage matrix is authoritative for each host and surface. hook status verifies configuration, not execution or delivery. Tool has not been independently tested. 🔹 numbat #AIagents #endpoint #cel #detection 🔗 Source: https://github.com/perplexityai/numbat

More than 30 Minnesota water systems were disrupted in one coordinated 48-hour attack. Hitting that many small utilities at once is the mark of one shared weakness, not 30 break-ins. The catchable signal sits on the IT servers, not the plant floor. https://suriq.io/blog/minnesota-water-utilities-coordinated-ot-attack #Detection #infosec #cybersecurity

Attackers pointed an off-the-shelf AI agent at Thailand's finance ministry and let it run the hack unattended. It used no new exploit. A data service left with no password got it in. Detect the host actions, not the AI. https://suriq.io/blog/hermes-ai-agent-thailand-finance-ministry #Detection #Linux #infosec #cybersecurity

RE: https://mstdn.social/@osnews/116965092743942475 Thank you to @onepict@chaos.social and @jwildeboer@social.wildeboer.net for this. My primary curiosity for digging in here wasn't the vendor or anti-vendor politics point, or the freedom-of-coder-communities-to-organize point, but rather… how exactly are they enforcing this? #AI #antislop #slop #detection #FOSS

Third on-prem SharePoint RCE this month: CVE-2026-50522 (CVSS 9.8) went from public PoC to active exploitation in hours. All three July flaws steal the server machine key. Patch, then ROTATE the key, or a patched box is still owned. #SharePoint #infosec https://suriq.io/blog/sharepoint-cve-2026-50522-rotate-machine-keys #CVE #Detection #CISAKEV #infosec

HollowGraph, an espionage implant, hides its command channel in a hijacked Microsoft 365 calendar over the Graph API. No CVE, no patch. Firewalls see normal M365 traffic. Hunt far-future calendar events with attachments in your audit logs. https://suriq.io/blog/hollowgraph-microsoft-365-calendar-c2 #CVE #ThreatIntel #Detection #infosec

Our weekly read: the three most dangerous advisories we tracked this week were not apps. They were the appliances you buy to stay safe. Dell Data Domain, Progress ShareFile and BeyondTrust all failed at authorization. Why that is the tier you watch least: https://suriq.io/blog/signal-safety-appliances-broke-authorization #Detection #infosec #cybersecurity

GigaWiper encrypts Windows files to a .candy extension with no key and no ransom note. That is on purpose: the ransomware is a decoy for a disk wiper. Microsoft and Binary Defense tracked the same code as one Iran-nexus operation. https://suriq.io/blog/gigawiper-fake-ransomware-disk-wiper #Ransomware #ThreatIntel #Detection #infosec

We beat Cloudflare's bot detection (open-source stealth browser) https://tilion.dev/blog/cloudflare-blocks-agents Comments: https://news.ycombinator.com/item?id=48867124 #HackerNews #Cloudflare #Bot #Detection #OpenSource #StealthBrowser #TechNews #CyberSecurity

Three separate July studies land on the same finding: AI coding agents can be made to run attacker code with the developer's own privileges. HalluSquatting, Friendly Fire, GhostApproval. No CVE, no patch. The fix is config plus detection, not a version bump. https://suriq.io/blog/hallusquatting-ai-coding-agent-endpoint-risk #SupplyChain #Detection #infosec #cybersecurity

Google Dialogflow CX had a flaw where one 'edit' permission on a chatbot ran code across every Code Block agent in the project, and reached cloud credentials. Google has fixed it. No CVE. The lesson: on managed AI platforms, an edit right is often a code-execution right. https://suriq.io/blog/dialogflow-cx-rogue-agent-edit-permission #CloudSecurity #Detection #infosec #cybersecurity

GodDamn ransomware loads PoisonX, a kernel driver reportedly carrying a valid Microsoft signature, to switch off EDR. Signature checks wave it through. Affects Windows servers. The tell is your agent going silent while still online. https://suriq.io/blog/signed-kernel-driver-edr-killer #Ransomware #Detection #infosec #cybersecurity

Cavern is a new Iran-linked backdoor that breaks into IT providers using no vulnerability at all. It abuses their own deployment tools, then hops to customers. No CVE to patch. Detection is the only defense. https://suriq.io/blog/cavern-manticore-c2-no-cve-it-provider-supply-chain #SupplyChain #Detection #infosec #cybersecurity

Kairos stole ~2TB from a US county, encrypted nothing, and was still paid $1M. No locker means no loud alarm. The only signals are the weak login in and the big upload out. Backups recover files, not a leaked copy. https://suriq.io/blog/kairos-encryptionless-extortion-detection #Ransomware #Detection #DataBreach #infosec

🚨 BREAKING FortiBleed, the campaign that scraped 110M logins from Fortinet firewalls, now has a buyer. SOCRadar tied it to INC and Lynx ransomware. One operator ran both gangs' panels; the access has caused 12 encryptions. Rotate Fortinet creds, hunt for backdoors. https://suriq.io/blog/fortibleed-credentials-inc-lynx-ransomware #Ransomware #CVE #CloudSecurity #Detection