Threat Radar — Top 10 Cyber Threats

CrowdStrike 2026: 89% YoY increase in AI-enabled attacks; eCrime breakout time 29 minutes (fastest: 27 seconds). 48% of security professionals identify agentic AI as top 2026 attack vector (Dark Reading poll). Microsoft April 2026: threat actor abuse of AI accelerating from tool to full cyberattack surface — adversaries weaponising agentic frameworks for autonomous recon, credential testing, and infrastructure rotation. Agentic attack breach window collapsed to 22 seconds (Jazz CyberShield, Apr 2026). First documented fully autonomous AI-orchestrated attack (Sep 2025): AI agent handled 80–90% of attack lifecycle without human control. Cascading multi-agent failure case: compromised procurement agent approved $3.2M in fraudulent orders before detection. ClawJacked vulnerability (Apr 9 2026): malicious websites silently hijack locally running OpenClaw agents via localhost WebSocket trust assumptions — brute-force gateway credentials from browser, gain administrative AI agent control (Barracuda, Apr 2026). CVE-2026-32211 (CVSS 9.1, Apr 3 2026): missing authentication in Microsoft Azure DevOps MCP framework enables unauthenticated code execution — first AI infrastructure CVE to reach critical severity in enterprise MCP tooling; confirms AI agent frameworks are now a primary attack surface (NIST NVD, Apr 2026). IBM Autonomous Security launched 15 Apr 2026: multi-agent defence service delivering coordinated decision-making at machine speed across full security stack — industry's first direct counter-capability to autonomous AI-driven attacks (IBM, Apr 2026).

US deepfake-related fraud losses tripled to $1.1B in 2025 (from $360M in 2024); projected $40B by 2027 (Deloitte). 72% of business leaders cite AI-enabled fraud as top operational challenge (Experian 2026). Experian warns of AI-powered emotionally intelligent bots sustaining dozens of simultaneous scam relationships. $25.6M Arup deepfake video-call heist remains landmark case. FBI reports AI voice cloning scams cost elderly Americans $2.3B in 2026 alone; global voice cloning losses projected $8B by year-end. Scam success rate climbed from 12% (2024) to 34% (2026) — voice clone possible from 60 seconds of audio. WEF March 2026: AI-fuelled cyber fraud now a global roadmap priority; UN News March 2026 global wake-up call — AI-fuelled organised fraud demands coordinated international response. Experian 2026 warns of AI-powered emotionally intelligent bots sustaining dozens of simultaneous scam relationships; 1-in-4 Americans now fooled by deepfake voice impersonation (UnboxFuture/Experian/UN, Apr 2026).

Five major supply-chain attacks in Q1 2026. North Korean actors spread 1,700+ malicious packages across npm, PyPI, Go, and Rust ecosystems (Hacker News, Apr 2026). Axios npm (100M weekly downloads) compromised by UNC1069 — WAVESHAPER.V2 backdoor deployed (Google GTIG, Apr 2026). TeamPCP multi-phase campaign affected 60+ npm packages between Feb–Mar 2026, hitting LiteLLM (3.4M daily PyPI downloads), Trivy, KICS, Telnyx — harvesting cloud credentials and CI/CD secrets via incomplete credential rotation (Datadog, Mar 2026). 36+ malicious npm packages exploited Redis/PostgreSQL for persistent implants. Attackers gaining trusted maintainer accounts rather than typosquats. Active Apr 2026: kube-health-tools campaign publishes simultaneous npm/PyPI packages deploying persistent reverse-tunnel implants on developer workstations and Kubernetes CI/CD pipelines (Panther/Zscaler ThreatLabz, Apr 2026).

Unit 42 2026 Global IR Report: attacks now 4× faster than prior year — fastest cases achieve initial access to full data exfiltration in 72 minutes. Over 7,500 organisations listed on leak sites in 2025 (+58% YoY); Q1 2026 total: 2,165 victims (annualises to 8,660 — +18.5% vs full-year 2025); record March 2026: 808 victims (highest monthly count ever, +33% above 2025 monthly avg); Qilin hits 131 in March alone — three consecutive months above 100, unprecedented for any group; 678 attacks reported Jan 2026 (+10% MoM); US attacks up 50% to 5,010 incidents (Unit 42/Recorded Future/Verizon 2026). Week of 11–19 Apr 2026: 185 victims recorded — April 13 alone saw 46 new victims announced in 24 hours; Qilin and DragonForce account for 21% of weekly volume (Ransom-DB, Apr 2026). 80% of ransomware attacks now leverage AI tools in some capacity. 57 new ransomware groups and 27 new extortion groups emerged in 2025. Median ransom paid jumped from $12.7K to $59.6K; mean recovery cost $1.53M; 64% of organisations refuse to pay. 87.6% of claims involve double extortion (encryption + exfiltration). Ransomware in 44% of all data breaches; 88% of SMB breaches. Top groups: Qilin, Akira, Clop, DragonForce, LockBit 5.0. Pure data extortion without encryption accelerating. Emsisoft Q1 2026 independent count: 2,318 incidents across 70 active ransomware groups — confirming record pace (StationX/BlackFog/VikingCloud/Emsisoft/Ransom-DB, Apr 2026).

Unit 42 2026: identity loopholes drive nearly 90% of all investigations. AiTM attacks increased 146% YoY with ~40,000 incidents detected daily. Starkiller phishing suite (Mar 2026) proxies real login pages via headless Chrome to bypass MFA in real time. Tycoon 2FA dismantled by Microsoft/Europol (early 2026) but successor platforms continue — accounted for 62% of phishing volume before takedown (30M+ fraudulent emails/month). >90% of credential compromises expected to involve automated phishing kits by end of 2026. Microsoft Security Blog (6 Apr 2026): AI-enabled device code phishing campaign targeting 340+ organisations globally — 10–15 campaigns every 24 hours since mid-March; AI-personalised lures by role, dynamic code generation defeating 15-minute expiry windows, EvilTokens PhaaS kit driving large-scale OAuth abuse; infrastructure hosted on Vercel/Cloudflare Workers/AWS Lambda to blend with legitimate traffic (Microsoft, Apr 2026). SpyCloud 2026 Identity Exposure Report: 65.7B identity records recaptured (+23% YoY) — largest identity intelligence datalake ever recorded; 8.6B stolen session cookies; 18.1M exposed API keys and tokens spanning cloud, AI, and payment platforms; 28.6M phished records (nearly half corporate users); successful phishing +400% YoY. 84% of compromised accounts had MFA enabled. Token theft dominant identity vector — AI-powered kits run real-time session hijacking (SpyCloud/WorkOS/Microsoft, Apr 2026).

Cloud misconfigurations cause 99% of security failures — avg 43 misconfigurations per account; 490% YoY spike in public SaaS attacks (CheckRed 2026). 100% of analysed companies operate SaaS environments with embedded AI; 80% of incidents involve PII/customer data. IBM X-Force 2026: cloud risk defined by identity exposure, weak admin practices, insecure integrations, and limited telemetry. Google Cloud Threat Horizons H1 2026: attackers deploying cryptominers in GKE instances within 1 hour of creation. Shadow AI added $670K to average breach cost; 50% of companies experienced AI-related data exposure. SaaS-to-SaaS OAuth chains create lateral movement invisible to perimeter controls (CheckRed/IBM/Google, Apr 2026).

Q-Day timeline accelerating sharply: three papers in three months have rewritten quantum resource estimates — what once required 20M qubits now potentially requires <100K under newer architectures (Quantum Insider, Mar 2026). Expert probability of cryptographically relevant QC in 10 years: 28–49% — highest ever recorded. 2026 designated "Year of Quantum Security" by FBI/NIST/CISA. Google sets 2029 internal PQC migration deadline; Pentagon 2030; UK NCSC three-phase: 2028/2031/2035; EU 18-nation statement targets CNI by 2030. >50% of web traffic through Cloudflare now uses PQ key agreement. NSA CNSA 2.0 mandates quantum-safe national security systems by Jan 2027. Canada mandates PQC plans from Apr 2026. World Quantum Day (14 Apr 2026): IBM, Google, and NIST researchers confirm state actors are actively stockpiling encrypted data under "Harvest Now, Decrypt Later" strategies at scale — PQC crypto inventory now considered an immediate board-level priority, not a future concern. Meta published PQC Migration Framework (16 Apr 2026) sharing production lessons across inventory, risk tiers, and deployment guardrails — Meta cryptographers co-authored NIST's HQC backup algorithm. US Senate panel advanced Quantum Initiative Reauthorization (15 Apr 2026) focusing on applications and security — signals accelerating federal quantum security policy (Quantum Insider/NCSC/NIST/Meta/World Quantum Day/US Senate, Apr 2026).

Ivanti EPMM zero-days CVE-2026-1281/1340 (CVSS 9.8) exploited since July 2025 — state-linked exploitation confirmed 6 months before disclosure; widespread exploitation began immediately post-patch. Fortinet FortiClient EMS CVE-2026-35616 (CVSS 9.1) actively exploited from 31 Mar 2026 — watchTowr detected exploitation 4 days before Fortinet advisory; CISA KEV catalog added 6 Apr; 2,000+ exposed instances (Shadowserver). Microsoft Patch Tuesday April 2026 (19 Apr 2026): 167 vulnerabilities patched including CVE-2026-32201 — actively exploited SharePoint Server zero-day enabling spoofing attacks that could expose or manipulate sensitive data; 20 remote code execution bugs rated Critical. Cisco (19 Apr 2026): four critical flaws in Identity Services and Webex enabling arbitrary code execution and user impersonation. Enterprise edge and endpoint software confirmed as highest-risk zero-day battleground (CyberNewsCentre, 9 Apr 2026). Window between disclosure and mass exploitation collapsed to hours. Edge devices from Barracuda, Citrix, Fortinet, Ivanti, Palo Alto, SonicWall under sustained nation-state and eCrime campaigns (Unit42/watchTowr/Tenable/Microsoft/Cisco, Apr 2026).

FBI confirms Salt Typhoon hacked 200+ companies across 80 countries (Aug 2025); Dec 2025: intrusions detected in US House committees. Volt Typhoon maintains 5+ year persistence in US energy, water, and transport CNI — rapidly rebuilt botnet after 2024 disruption. Feb 2026: Senator Cantwell demands AT&T/Verizon CEO testimony; Mandiant assessment reports still withheld. US House Oversight hearing concludes federal agencies need "proactive cybersecurity strategy" against state-sponsored threats. FBI IC3 2025 report: US cybercrime losses hit $21B — CNI threats intensifying. CISA counter-advisory AA25-239A targets Chinese actors worldwide. Finland flags Russian/Chinese cyber espionage targeting government and CNI. BRICKSTORM malware actively deployed against VMware vSphere. Waterfall Threat Report 2026: ransomware slowdown masks deeper shift toward nation-state attacks on physical processes in OT/CNI environments; Finland National Security Overview 2026: Russian/Chinese espionage targeting government and CNI flagged as top national security threat Norwegian intelligence (PST/NSM, Apr 2026) confirmed Salt Typhoon as part of a broader campaign against NATO allies. DHS disclosed Salt Typhoon breached US Army National Guard infrastructure, exposing admin credentials and network diagrams — first confirmed military affiliate breach. CISA advisory AA26-097a (7 Apr 2026): Iranian-affiliated APT actively targeting internet-exposed PLCs across US water/wastewater and CNI sectors — 75+ Unitronics HMI devices compromised; FBI assesses escalation linked to Iran-US-Israel hostilities. Iran-linked Handala group claimed attack on Stryker Corp (11 Mar 2026) disrupting manufacturing, order processing, and shipping via Microsoft systems. CERT-UA (Apr 2026): UAC-0247 campaign targeting Ukrainian government agencies and municipal healthcare institutions (clinics and emergency hospitals) Mar–Apr 2026 — delivering data-stealing malware harvesting Chromium browser data and WhatsApp communications; campaign attributed to state-backed threat cluster (DHS/CISA/Palo Alto Unit42/Recorded Future/CERT-UA, Apr 2026).

Insider risk costs hit $19.5M per organisation annually — up 123% since 2018 (Proofpoint 2026). Only 10% report zero incidents (down from 17%); 20+ incidents per year doubled. 60% of organisations express high concern over AI-amplified insider risk; 73% of IT staff say AI creates invisible exfiltration paths. Fastest data exfiltration cases in 2026 occur 4× quicker than prior year — one intrusion achieved exfiltration within 4 minutes of access. 39.7% of all AI interactions involve sensitive data, often unintentionally (Cyberhaven Labs). Only 20% confident they can detect AI-related insider incident before significant damage. Gurucul 2026: "AI became an insider" — shadow AI added $670K to avg breach cost. Thales 2026 Data Threat Report: AI now primary emerging insider threat vector; 61% cite AI as #1 data security risk (new all-time high, up from 51% in 2025); 47% of sensitive cloud data remains unencrypted; 70% of IT/security professionals say AI transformation pace is their most significant security challenge; only 34% of organisations know where all their data resides. IBM X-Force 2026: infostealer malware exposed 300,000+ ChatGPT credentials in 2025 — AI platforms now equal credential risk to core enterprise SaaS (IBM/Proofpoint/Thales/Cyberhaven, Apr 2026).