PamStealer now needs its C2 server to unlock the macOS payload.

The new variant uses an X25519 key exchange, a fake Wavel wallet lure, and redundant persistence, including global Git hooks.

Read: https://thehackernews.com/2026/09/pamstealer-macos-malware-adds-live-c2.html

‼️ Mini Shai-Hulud came back without a new attacker update.

Two compromised GitHub Actions became reachable again while their tags still pointed to malicious commits, letting downstream workflows resume executing the credential stealer.

🔗 Here's how the attack reactivated: https://thehackernews.com/2026/09/compromised-github-actions-came-back.html

🚨 Kiteworks tells customers to shut systems down for nine hours after federal intelligence warned that a threat actor may target some of its systems.

The company says it has found no evidence of compromise and recommends upgrading to version 9.5.1.

Read: https://thehackernews.com/2026/09/kiteworks-urges-customers-to-shut-down.html

‼️ Attackers are exploiting a SharePoint flaw that Microsoft says can enable remote code execution.

CISA also flagged a MikroTik RouterOS flaw used in a chain that gives unauthenticated attackers full admin access to vulnerable routers.

Learn how the two attack paths work: https://thehackernews.com/2026/09/sharepoint-rce-and-mikrotik-routeros.html

⚠️ Elementor CSRF flaw can create a rogue WordPress admin account when an administrator opens a crafted link.

The bug affects 4.3.0 and 4.3.1, bypasses CSRF protection across the site's REST API, and is fixed in 4.3.2.

Here's how it works: https://thehackernews.com/2026/09/elementor-csrf-flaw-lets-attackers-take.html

🚨 An attacker bypassed auth on an employee-run AI agent, obtained its model provider API key, and racked up the equivalent of $600,000 in token use over three weeks.

There was no spending limit, and dashboard gaps helped the activity go unnoticed.

🔗Why visibility comes first → https://thehackernews.com/2026/09/zero-trust-for-ai-agents-starts-with.html

🚨 One URL-encoded character is helping attackers bypass WAF rules protecting Oracle PeopleSoft.

UNC6240/ShinyHunters is exploiting CVE-2026-35273 to deploy web shells on dozens of systems, then using post-exploitation tools including SIDEEYE.

Read: https://thehackernews.com/2026/09/attackers-bypass-wafs-to-exploit-oracle.html

🚨 Lunex uses a vulnerable AMD driver to blind security monitoring before stealing browser credentials.

The BYOVD chain zeroes kernel callbacks tied to security products, then the stealer collects credentials and cryptocurrency wallet data.

🔗 Read more → https://thehackernews.com/2026/09/lunex-stealer-abuses-amd-driver-to.html

‼️ WARNING — Citrix NetScaler is facing two unpatched RCE 0-days under active exploitation.

Researchers at watchTowr say the flaws are being exploited in the wild. Citrix has not confirmed them or released a patch, affected-version guidance, workaround, or indicators of compromise.

What defenders should know now: https://thehackernews.com/2026/09/warning-two-unpatched-citrix-netscaler.html

‼️ BREAKING: Citrix confirms two NetScaler flaws have been exploited and has released fixes. CVE-2026-88771 and CVE-2026-88772 were observed exploited on unmitigated deployments.

The new advisory covers 8 new CVEs affecting NetScaler ADC and Gateway.

Read details → https://thehackernews.com/2026/09/warning-two-unpatched-citrix-netscaler.html

⚠️ CISA says attackers are actively exploiting two critical Citrix NetScaler flaws globally.

CVE-2026-88771 can allow unauthenticated command execution. CVE-2026-88772 can enable RCE when DTLS is enabled.

Citrix has fixes and IoCs.

Read: https://thehackernews.com/2026/09/cisa-says-attackers-are-exploiting-two.html

🚨 JADEPUFFER deleted most targeted Azure Storage accounts in seven minutes.

Microsoft says compromised service principals drove 150+ destructive or credential-collection operations in 35 minutes, while the attack also targeted backup and recovery resources.

Inside the 18-hour attack: https://thehackernews.com/2026/09/jadepuffer-linked-attackers-used.html

⚠️ Carbonato compromises unauthenticated Docker daemons, then deploys Hermes Agent for Telegram-driven, LLM-generated commands.

The botnet persists on infected hosts and scans nearby networks every five minutes for more exposed Docker daemons.

Here's how it works: https://thehackernews.com/2026/09/carbonato-botnet-compromises-docker.html

⚡ WEBINAR — If an AI agent has access to your business systems, who approved it, what can it reach, and does it still need those permissions?

Only 47% of CISOs say they can identify every AI agent in their environment.

Join us to learn how to treat AI agents as first-class identities, with clear ownership, tighter permissions, regular access reviews, and faster revocation when risk changes.

Watch the webinar → https://thehackernews.com/2026/09/webinar-how-to-govern-ai-agents-reduce.html

⚡ THIS WEEK IN CYBER...

> $387M crypto hack
> Citrix flaws exploited
> AI agents hacking sites
> 1.7K-repo domain trap
> 5.7K M365 accounts targeted
> EvilTokens taken down
> TeamCity abused in ransomware
> Git-based C2 campaigns
> Clop leak site hacked
> PamStealer gets stealthier

and dozens more CVEs, campaigns, and attacks.

Full weekly recap → https://thehackernews.com/2026/09/weekly-recap-387m-crypto-hack-citrix.html

AI raised engineering velocity for 93% and governance workload for 97%.

ActiveState's new report, The AI Remediation Bill, surveyed 307 enterprise security, governance, platform, and engineering leaders and found that 93% report increased engineering productivity from AI-assisted development, while 97% report a heavier governance workload.

Join the waitlist → https://thn.news/ai-report-syndicate

⚠️ RatHat, an Android banking trojan, uses Gemini to rank infected phones by estimated bank balance.

Cleafy traced nearly 100 console deployments since April 2026. Gemini helps operators prioritize victims, not move money.

Read: https://thehackernews.com/2026/09/rathat-android-malware-console-uses.html

🛑 Bitget says an attacker exploited a third-party security product flaw, stole high-level internal credentials, and used them in a $388M theft.

The wallet system accepted fraudulent withdrawal commands as legitimate, bypassing risk controls.

Read: https://thehackernews.com/2026/09/bitget-says-attacker-exploited-third.html

🚨 IAM can show what an AI agent is allowed to do. It may miss what the agent actually does.

Agents can chain tools and actions inside applications that authentication logs may not capture. That makes runtime telemetry, scoped delegation, and distinct agent identities part of the control model.

Why traditional IAM falls short: https://thehackernews.com/2026/09/iam-for-ai-agent.html

⚠️ NeedyMantis keeps attackers inside networks they already breached.

Microsoft saw the malware in a small number of targeted intrusions, where it uses DLL sideloading and a WebSocket C2 channel to load additional modules.

Read: https://thehackernews.com/2026/09/hackers-use-needymantis-to-maintain.html