A leaked credential is only useful intelligence if you know it still works.

Pentera’s Recorded Future integration automatically tests exposed credentials against an organization’s external attack surface to confirm which ones can actually be used.

How the validation works → https://thehackernews.com/2026/09/threat-intelligence-alone-wont-close.html

🛑 A Parallels Desktop flaw lets non-admin Mac users gain root, but Intel Macs can’t install the version containing the fix.

Tracked as CVE-2026-90894, JFrog demonstrated the flaw on Parallels Desktop 26.4.0 and reports no attacks.

How the root escalation works → https://thehackernews.com/2026/09/parallels-desktop-flaw-lets-non-admin.html

⚠️ N0va phishing abuses legitimate sign-in flows to capture access and refresh tokens.

The campaign impersonates Teams, SharePoint, DocuSign, Google Drive, and other trusted services, then can establish SSO access to email, files, and cloud apps.

Inside the attack chain → https://thehackernews.com/2026/09/n0va-phishkit-targets-us-and-eu.html

⚠️ One hijacked AI coding session helped spread Shai-Hulud across about 100 internal repositories.

A poisoned PyPI package installed an infostealer, GitHub OAuth tokens were stolen, and a compromised package in the company’s own namespace caused a second infection.

Read: https://thehackernews.com/2026/09/attacker-hijacks-ai-coding-assistant.html

‼️ A browser extension could hijack AI agents in Comet, Edge, Opera Neon and Claude, while abusing Chrome’s Gemini integration to reach local files, the camera and mic.

Researchers used the same underlying trust-boundary weakness across all five products after the extension was already installed.

Here's the technique → https://thehackernews.com/2026/09/one-extension-could-hijack-ai.html

🚨 From compromised VPN credentials to MQTT and Matrix C2, three threat groups are targeting Russian enterprises.

NightEagle uses GhostContainer, Toy Ghouls now fields a custom backdoor, and Kaspersky links Hacking Cat to Gorilla RAT and destructive malware.

Read: https://thehackernews.com/2026/09/three-threat-groups-target-russian.html

‼️ Attackers are exploiting a critical Issabel Framework flaw.

CVE-2026-89026 uses a hard-coded JWT signing key, letting unauthenticated remote attackers forge tokens and execute OS commands as the Asterisk user. A fix is available.

How the flaw works: https://thehackernews.com/2026/09/attackers-exploit-issabel-framework.html

⚡ U.S. seizes two NightmareStresser domains tied to hundreds of thousands of DDoS attacks and attempts.

The DDoS-for-hire service was assessed to have been used worldwide since 2022.

Operation PowerOFF details → https://thehackernews.com/2026/09/us-seizes-nightmarestresser-domains.html

‼️ WARNING - New Cisco ISE flaw CVE-2026-76460 is under active exploitation.

The CVSS 10.0 auth bypass has no workaround, and successful exploitation may lead to root-level command execution.

Affected versions, fixes, and compromise checks: https://thehackernews.com/2026/09/cisco-warns-of-new-zero-day-ise-auth.html

⚠️ Gyazo breach exposed 23.62 million user records and 490 million image metadata records.

The leaked data included IDs used to build Gyazo image links. Helpfeel says those IDs could allow unauthorized viewing of some images.

Read: https://thehackernews.com/2026/09/gyazo-breach-exposes-2362-million-user.html

🚨 A single unauthenticated DoH request can BIND’s named process under a specific connection-close condition.

ISC patched 14 BIND flaws, including crashes, resource exhaustion, and cache-poisoning issues. No active exploits are known.

Details: https://thehackernews.com/2026/09/bind-9-update-fixes-14-flaws-including.html

‼️ An OpenAI model found and used an exposed GitHub API key without authorization.

The key worked, but the requested data remained unavailable. The model then fabricated the data and attributed it to the requested source.

https://thehackernews.com/2026/09/openai-reveals-six-model-incidents.html

It’s one of 6 misalignment incidents OpenAI disclosed.

Attackers weaponize new vulnerabilities in ~5 days. The median org takes 43 days to patch known-exploited flaws.

Peer-reviewed AI agents exploited 87% of one-day vulnerabilities unaided.

The real test: provable coverage, independent validation, and safe-stop controls in production.

https://thehackernews.com/2026/09/cisos-expert-guide-to-agentic.html

🛑 China-aligned FamousSparrow has shifted to a previously unreported backdoor in attacks across Latin America.

SparroWocky can run commands, exfiltrate files, take screenshots, and load in-memory plugins, with government entities among the targets.

Read: https://thehackernews.com/2026/09/china-aligned-famoussparrow-deploys.html

Meet Georgetown's cybersecurity faculty on Sept 22 to learn more about the Cybersecurity Risk Management master's program.

Attend our virtual webinar on Sept 22.

Learn more → https://thn.news/risk-management-event

‼️ Critical Unbound DNSSEC validator flaw (CVE-2026-82717) could allow remote code execution.

An attacker who controls a malicious zone can trigger the heap overflow by querying a vulnerable resolver. 1.26.0 and earlier are affected.

Read details → https://thehackernews.com/2026/09/critical-unbound-dnssec-validator-flaw.html

⚡ Can you prove a CVE is exploitable without running the exploit?

This webinar shows how to map a vulnerability to the attack techniques it requires, then test whether your existing controls actually stop them.

Can’t join live? Sign up and get the recording: https://thehacker.news/ready-for-mythos-class-attacks

🚨 Iran-linked Handala Hack is tied to a Telegram-controlled backdoor that can steal passwords and messaging data.

HEAVYGRAM also executes commands, captures screenshots, and downloads additional malware.

See how Telegram becomes the control channel: https://thehackernews.com/2026/09/iran-linked-handala-hack-tied-to.html

‼️ ALERT - Critical Docker Sandboxes flaw lets malicious guest code escape the shared workspace and read or modify files across a macOS host.

CVE-2026-77179 crosses the virtio-fs boundary with the host account’s rights.

Read how the escape works → https://thehackernews.com/2026/09/critical-docker-sandboxes-flaw-lets.html

🔥 25 New CYBER stories in this week’s ThreatsDay Bulletin:

→ Exposed AI systems breached
→ Stealers hunt agent tokens + prompts
→ Agents rewrite their own models
→ Ransomware jumps on a fresh flaw
→ Telnet brute-force still works
→ RF attacks leak audio remotely
→ and more...

Check out the full list of threats here: https://thehackernews.com/2026/09/threatsday-self-rewriting-agents-800.html

‼️ NEW - Critical Check Point management flaw lets unauthenticated attackers run code as root.

The 9.8-rated bug (CVE-2026-91843) sits in the login path before authentication.

Affected versions and fix details: https://thehackernews.com/2026/09/critical-check-point-management-server.html