⚡ The U.S. has ordered federal civilian agencies to remove DeepSeek products from their IT systems and stop buying from the Chinese AI company.

The DHS order also covers third-party products and services built using DeepSeek implementations.

Details: https://thehackernews.com/2026/10/us-orders-deepseek-removed-from-federal.html

🚨 WordPress sites backdoored using valid UpdraftCentral keys.

Attackers pushed malicious plugins and created rogue admin accounts after TeamUpdraft disclosed a website breach.

99 distinct backdoor plugin names observed.

Details: https://thehackernews.com/2026/10/attackers-abuse-updraftcentral-keys-to.html

🛑 Attackers hijacked over 300 Tor exit relays to intercept SSH connections and steal login passwords.

A researcher found 333 exits redirecting SSH traffic to 12 interception servers.

The Tor Project has since blocked the affected relays.

Here's how the attack worked: https://thehackernews.com/2026/10/attackers-hijacked-over-300-tor-exit.html

🚨 PHP2Shell: the one-line “fix” everyone shared overnight — session.upload_progress.enabled=0 — isn’t enough on its own, the firm that suggested it now says.

A PHP core developer put it bluntly: “the only mitigation is patching / upgrading PHP.”

No patch until 14:00 UTC today.

Here’s what actually cuts your risk before then → https://thehackernews.com/2026/10/php2shell-what-actually-protects-you.html

AI agents can find web app vulnerabilities. But can they prove the exploits without crossing security boundaries?

Pentera Apps pairs AI-driven pentesting with a deterministic harness that controls execution, logs actions, and generates replayable exploit scripts.

Here's how it works: https://thehackernews.com/2026/10/the-harness-is-hard-part-of-agentic.html

🚨 30+ real estate agents. Different senders. Unique property inquiries. One phishing objective.

Proofpoint reveals how an attacker used personalized conversations and video meeting invites, including a link abusing Google's redirect service, to disguise phishing attempts.

Here's how the campaign worked: https://thehackernews.com/2026/10/generative-ai-is-erasing-behavioral.html

🛑 China-aligned threat group Bronze Sandcastle is targeting organizations in Asia and Africa, abusing Sandboxie DLL sideloading to deploy encrypted backdoors.

Sophos reveals its attack methods, payloads, and infrastructure.

Read more → https://thehackernews.com/2026/10/china-aligned-bronze-sandcastle-abuses.html

🚨 A trusted Adobe link could be the first step to an enterprise breach.

GeorgeGinx PhaaS is targeting U.S. organizations with device code phishing, exploiting familiar services to deceive users. 👀

ANY.RUN TI Feeds equip SOCs and MSSPs with unique threat indicators to identify malicious infrastructure and strengthen detection.

📊 Nearly 1 in 3 unique malicious IOCs found in sandbox investigations were already covered by ANY.RUN TI Feeds.

👉 Explore TI Feeds and give your team a smarter way to stay ahead of emerging threats: https://thn.news/live-threat-feeds-tl

🔥 Google Chrome is rolling out new protections against #ClickFix attacks that trick users into running malicious commands.

These attacks bypass browser download protections. Chrome now detects suspicious copied commands and warns Enhanced Protection users before execution.

Read more: https://thehackernews.com/2026/10/google-chrome-adds-clickfix-warnings-to.html

‼️ BREAKING - PHP releases emergency patches for two critical #PHP2Shell RCE flaws (CVSS 9.2) as early exploit attempts emerge.

The flaws can enable unauthenticated code execution and expose server memory.

👉 PATCH NOW (CVE-2026-109618, CVE-2026-109619)

Details ➜ https://thehackernews.com/2026/10/php-issues-emergency-patches-for-two.html

Every security vendor is shipping AI agents. Few are asking what those agents are actually acting on. 🤔

An agent that remediates based on one tool’s partial view of an asset isn’t saving time. It’s automating mistakes at machine speed.

The secret weapon isn’t the AI. It’s the data underneath it.

That’s the bet Axonius is making with its AI Workforce: Virtual Workers that monitor, flag, and fix, all grounded in one reconciled view of every asset and exposure. Humans stay in the loop where it matters, and autonomy is earned, not assumed.

Is your AI working from ground truth, or a best guess? 👇

Learn more: https://thn.news/ai-threat-defense

⚠️ OpenAI's GPT-6 agents disabled security alerts and removed safety monitoring while carrying out routine coding tasks.

In one case, an agent falsely denied removing monitoring functionality.

OpenAI is now investigating why its agents weakened safeguards meant to oversee their behavior.

Read - https://thehackernews.com/2026/10/openais-gpt-6-agent-removed-safety.html

🚨 LiteLLM 1.107.0 compromised in a supply chain attack.

Malicious PyPI packages and Docker images delivered a credential stealer and backdoor.

⚠️ The malware runs at Python startup, targeting cloud credentials, private keys, and AI tool credentials.

🔗 Full story: https://thehackernews.com/2026/10/litellm-compromised-in-supply-chain.html

🚨 CRITICAL: GitLab flaw (CVSS 10.0) affects 12 years of releases.

CVE-2026-110264 could allow unauthenticated remote code execution on self-managed servers.

Read - https://thehackernews.com/2026/10/critical-gitlab-flaw-cvss-100-could.html

🚨 China-linked hackers used ARTEX AI agents to abuse stolen ConnectWise app credentials and gain access to customer cloud environments.

Microsoft ties Storm-0940 to the ConnectWise Cloud Backup supply-chain compromise.

Details ↓ https://thehackernews.com/2026/10/china-linked-hackers-use-artex-ai.html

‼️ ALERT - Someone tried to exploit #PHP2Shell 4 days before the vulnerabilities were publicly disclosed.

Amazon caught the October 17 attempt. More exploit attempts followed the patch release.

Who had the exploit details early, and how?

Full story ↓ https://thehackernews.com/2026/10/someone-tried-to-exploit-php2shell-4.html

‼️ A single HTTP/3 connection can crash vulnerable web servers. No login required.

Newly disclosed QPACK flaws trigger up to 4,000× memory amplification in tests, exhausting server resources.

→ PoC is public
→ 7 implementations affected
→ Still unpatched: Bun, aioquic, h3

Full details ↓ https://thehackernews.com/2026/10/new-qpack-flaws-can-crash-http3-servers.html

🚨 Node.js patches 10 security flaws, including a high-severity Brotli decoder bug that could allow RCE when apps decompress attacker-controlled data.

Patches are out for versions 22.x, 24.x, and 26.x.

Update Now ↓ https://thehackernews.com/2026/10/nodejs-brotli-decoder-bug-could-allow.html

⚠️ LiteLLM’s release pipeline was hijacked.

Malicious v1.107.0 packages shipped a credential stealer and backdoor. Attackers later tested stolen AWS credentials.

LiteLLM suspects GitHub OAuth tokens were taken from logs on its exposed internal deployment.

Read ↓ https://thehackernews.com/2026/10/litellm-suspects-github-token-leak-in.html

🚨 OpenCode flaw lets malicious MCP servers execute code on Windows and WSL when developers initiate sign-in.

Fixed in v1.18.33, but there's no CVE or security advisory. Vulnerability scanners won't flag affected versions.

What developers need to know: https://thehackernews.com/2026/10/opencode-flaw-lets-malicious-mcp.html