Machine identities now outnumber humans 82 to 1.

AI attackers use that sprawl to hop from dev containers to production systems in seconds, chaining minor gaps into real attack paths. Critical flaws aren’t required.

Patch cycles can’t match machine speed.

🔗 AI-driven breach paths + CTEM response → https://thehackernews.com/2026/02/from-exposure-to-exploitation-how-ai.html

The SANS State of ICS/OT Security 2025 Report reveals an industry advancing at two speeds.

Detection is faster, but recovery still lags—with one in five incidents taking over a month to restore operations.

Get the intel 👉 https://thn.news/ics-security-25-x

⚠️ Android malware now uses Gemini AI.

Researchers say PromptSpy sends screen XML to Gemini, gets JSON tap steps, and pins itself in Recent Apps to stay alive.

It then enables VNC remote access.

🔗 Learn how the AI loop keeps it alive → https://thehackernews.com/2026/02/promptspy-android-malware-abuses-google.html

🛑 Microsoft patched a Windows Admin Center flaw enabling privilege escalation across managed systems.

CVE-2026-26119 (CVSS 8.8) stems from improper authentication and could grant rights equal to the running user.

🔗 Read → https://thehackernews.com/2026/02/microsoft-patches-cve-2026-26119.html

651 arrests across 16 African nations after an INTERPOL crackdown on online scams.

Operation Red Card 2.0 hit investment fraud, mobile loan apps, and telecom breaches. Investigators linked cases to $45M in victim losses and seized thousands of devices.

🔗 Read → https://thehackernews.com/2026/02/interpol-operation-red-card-20-arrests.html

This week’s ThreatsDay recap starts with core platform risk.

🔓 OpenSSL RCE flaws
📄 Foxit PDF exploits
🤖 Copilot DLP bypass
📬 Enterprise email exposure
🛠 Patches issued
🧪 ClickFix macOS hits
📦 Loaders → RATs/ransomware
🌐 Typosquat delivery chains
🏭 119 groups hit OT/ICS
🔗 Supply-chain entry tactics
🔑 LLM-made passwords
📊 Weak randomness patterns
🎯 Predictable outputs
🛡 Security workflow risk

🔗 Complete recap → https://thehackernews.com/2026/02/threatsday-bulletin-openssl-rce-foxit-0.html

Three former tech employees were indicted for allegedly stealing Google trade secrets and transferring them to Iran.

Prosecutors say the data involved Pixel Tensor processor security and cryptography. Files were copied, photographed, and concealed.

🔗 Details → https://thehackernews.com/2026/02/three-former-google-engineers-indicted.html

The FBI warns ATM “jackpotting” caused over $20M in losses in 2025.

Since 2020, 1,900 incidents have been reported, including 700 last year. Attackers use #malware like Ploutus to bypass bank authorization via the XFS layer & trigger rapid cash-outs.

🔗 Read → https://thehackernews.com/2026/02/fbi-reports-1900-atm-jackpotting.html

A 29-year-old Ukrainian was sentenced to prison for aiding North Korea’s IT job fraud scheme.

He admitted selling stolen U.S. identities through a site seized in 2024, helping overseas workers secure jobs.

🔗 Details → https://thehackernews.com/2026/02/ukrainian-national-sentenced-to-5-years.html

🛡️ One in three cyber-attacks starts with a compromised employee account, pushing insurers to 🔐 audit passwords, admin access, and full MFA enforcement.

Coverage now depends on proving identity risk is tightly controlled.

🔗 Why MFA gaps can cost millions → https://thehackernews.com/2026/02/identity-cyber-scores-new-metric.html

MIMICRAT, a new RAT, is spreading via compromised legitimate sites.

Hijacked services displayed fake Cloudflare checks, pushing a PowerShell command that disables logging and AV, then connects over HTTPS masked as analytics traffic.

🔗 Loader stages and 22-command toolkit → https://thehackernews.com/2026/02/clickfix-campaign-abuses-compromised.html

⚠️ WARNING: Cline CLI was silently altered for 8 hours after a stolen npm token was used to publish v2.3.0 with a hidden postinstall script that installed OpenClaw.

Roughly 4,000 downloads occurred before the release was pulled & the token revoked.

🔗 Read → https://thehackernews.com/2026/02/cline-cli-230-supply-chain-attack.html

⚠️ A 9.9-rated Roundcube RCE flaw is now on CISA’s KEV list after confirmed active exploitation.

Researchers say attackers diffed and weaponized the bug within 48 hours. An exploit was reportedly offered for sale in June 2025.

The issue lived in the code for more than 10 years.

A second XSS flaw is also being abused.

🔗 Read → https://thehackernews.com/2026/02/cisa-adds-two-actively-exploited.html

🔥 Anthropic is rolling out Claude Code Security, an AI tool that scans full codebases and suggests patches.

In limited preview for Enterprise and Team users, it analyzes code like a human, traces data flows, and reduces false positives in a review dashboard.

All fixes require human approval.

🔗 Learn how it works → https://thehackernews.com/2026/02/anthropic-launches-claude-code-security.html

Generative AI traffic is up 890%, and 87% of organizations report AI-driven attacks.

EC-Council has launched four AI certifications plus Certified CISO v4 to help teams handle adoption, security, and governance as AI risk grows.

🔗 Framework and certification details → https://thehackernews.com/2026/02/ec-council-expands-ai-certification.html

🚨 A financially driven actor breached 600+ FortiGate devices across 55 countries using commercial AI.

No zero-days. They scanned exposed management ports, brute-forced weak logins, accessed VPNs, ran DCSync in AD, and targeted Veeam backups.

🔗 Details here → https://thehackernews.com/2026/02/ai-assisted-threat-actor-compromises.html

⚡ Android 17 beta blocks cleartext traffic by default.

Apps targeting 17+ must define a Network Security Configuration; usesCleartextTraffic="true" alone won’t allow HTTP. Google also adds HPKE hybrid cryptography for stronger app communications.

🔗 Read → https://thehackernews.com/2026/02/threatsday-bulletin-openssl-rce-foxit-0.html#privacy-model-hardening

Iran’s MuddyWater launched Operation Olalampo on Jan 26, 2026, targeting organizations across MENA.

Group-IB says phishing Office macros drop new malware—GhostFetch, GhostBackDoor, HTTP_VIP, and the Rust backdoor CHAR.

Some variants use Telegram for control, with signs of AI-assisted development.

🔗 Read → https://thehackernews.com/2026/02/muddywater-targets-mena-organizations.html

Microsoft says a Copilot bug (CW1226324) let Microsoft 365 Copilot summarize confidential emails, bypassing DLP policies.

Since Jan 21, 2026, emails in Sent Items and Drafts with sensitivity labels were processed in Copilot Chat without permission.

Microsoft fixed the issue on Feb 3 but hasn’t disclosed impact.

🔗 Details → https://thehackernews.com/2026/02/threatsday-bulletin-openssl-rce-foxit-0.html#copilot-bypassed-dlp-safeguards

🤖 Researchers found 19 malicious npm packages spreading SANDWORM_MODE.

The worm 🐛 steals npm/GitHub tokens, SSH keys, API secrets, and crypto keys, then propagates using stolen identities. It also injects into AI coding tools to harvest LLM API keys.

🔗 Read → https://thehackernews.com/2026/02/malicious-npm-packages-harvest-crypto.html

⚠️ The real risk in enterprise AI isn’t the model — it’s the endpoint.

Every new LLM API, dashboard, or connector expands the attack surface. Many were built fast, not secure. Exposed endpoints can leak data or inherit powerful service account access.

🔗 Why endpoint privilege now matters in LLM stacks → https://thehackernews.com/2026/02/how-exposed-endpoints-increase-risk.html