Daily Cybersecurity News – November 2, 2026

What happened

Security researchers at Anthropic identified a critical vulnerability in the Linux kernel that allows unauthenticated remote code execution on systems using Multipath TCP (MPTCP) listeners.

The flaw enables attackers to achieve RCE without authentication by targeting MPTCP-enabled systems. Further details and coverage appeared in Bleeping Computer reports.

Who is affected

Linux systems configured with Multipath TCP listeners. This includes servers and devices running vulnerable kernel versions that have MPTCP enabled, potentially a significant portion of modern Linux deployments where the feature is active.

Why it matters

Unauthenticated RCE in the kernel represents one of the most severe classes of flaws, allowing complete system compromise from remote networks without credentials. Operators running MPTCP face immediate risk of full takeover, data theft, or further lateral movement.

How it could have been prevented

Disable MPTCP if not required. Apply available kernel patches promptly once released by distributors. Monitor for unusual network activity on MPTCP ports and restrict exposure of listeners to trusted networks only.

What happened

The group behind the Aur0ra ransomware, tracked by Sophos as GOLD OGRE, has been using AI agents to automate much of its post-compromise activity inside victim networks.

Attackers supplied the agents (via Cursor AI coding assistant and models from Anthropic, xAI, and Moonshot) with VPN logins, proxy access, targets, and objectives. The agents then mapped networks and Active Directory, hunted for passwords in shares and backups, and moved laterally, sometimes reaching domain admin. Breaking in and deploying ransomware remained under human control. Prompts were often written in Russian, and operators framed activity as authorized red-team work to bypass guardrails.

Who is affected

Organizations hit by Aur0ra ransomware operations. Sophos recovered artifacts showing successful domain management access in some cases, with three victims later appearing on the group's leak site. Other ransomware groups are expected to adopt similar AI agent techniques.

Why it matters

AI agents lower the skill bar and accelerate credential hunting, lateral movement, and privilege escalation, allowing fewer operators to handle more intrusions. Defenders must assume faster attacker progress once initial access is gained and prioritize rapid detection and response.

How it could have been prevented

Patch internet-facing devices promptly. Deploy phishing-resistant multi-factor authentication (MFA). Monitor EDR alerts closely and ensure quick incident response. Restrict and audit use of AI coding assistants in sensitive environments; watch for unusual agent activity or proxy setups.

What happened

Attackers posing as IT support on Microsoft Teams socially engineered employees into granting remote control of their PCs via Windows Quick Assist. They then abused insecure Active Directory Certificate Services (AD CS) configurations to obtain login certificates for privileged accounts, including domain controllers.

Microsoft investigated two such intrusions. No vulnerability in Teams itself was exploited. In one case, data was copied to an external server. Attackers used ESC1-weak certificate templates allowing low-privileged users to request certs impersonating high-privileged accounts. Certificates persist even after password resets. One intrusion involved DCSync after gaining a domain controller cert.

Who is affected

Organizations using Microsoft Teams and Active Directory with poorly configured AD CS certificate templates (especially ESC1). Employees who can be reached via Teams from external tenants. Two investigated cases affected multiple users and privileged accounts.

Why it matters

Social engineering plus durable certificate-based access bypasses many traditional containment steps like password resets or machine wipes. Privileged certs enable domain dominance and credential dumping, leading to broad data theft and persistence that is hard to eradicate.

How it could have been prevented

Harden AD CS: audit and remediate ESC1 and similar vulnerable certificate templates so requesters cannot specify arbitrary subject names for privileged accounts. Train users to verify IT support contacts out-of-band and never share Quick Assist codes with unsolicited callers. Enable and tune automatic attack disruption in Microsoft Defender. Restrict external Teams federation where possible and monitor for unusual certificate enrollments and DCSync activity.

What happened

Ivanti released patches for two additional Connect Secure zero-days, CVE-2026-111357 and CVE-2026-118527, that attackers chained with the earlier CVE-2026-104935 authentication bypass in zero-day attacks.

CISA added both new CVEs to its Known Exploited Vulnerabilities catalog. CVE-2026-111357 (CVSS high) is a privilege escalation due to mishandled trust of DSCache configuration data, allowing elevation to root. CVE-2026-118527 involves predictable identifiers enabling authenticated file uploads to known locations. Attacks hit a limited number of customers. Fixes ship in ICS 25.1.1.3 and 22.8R2.6.

Who is affected

Users of Ivanti Connect Secure versions 25.1.1.2 and prior, and 22.8R2.5 and prior. Organizations that only applied the prior patch for CVE-2026-104935 must upgrade again. Limited confirmed victims so far, but VPN appliances are high-value targets.

Why it matters

Chained zero-days on internet-facing VPN gateways give attackers a reliable path from unauthenticated access to full device compromise and network foothold. Active exploitation confirmed by CISA means unpatched systems are at immediate risk of breach.

How it could have been prevented

Upgrade immediately to Ivanti Connect Secure 25.1.1.3 or 22.8R2.6 (or later). Follow CISA KEV guidance and apply patches on the mandated timeline. Monitor for indicators of compromise from the related advisories, restrict management interfaces, and consider compensating controls such as additional authentication layers or network segmentation for VPN traffic.

What happened

The FBI and U.S. Secret Service attributed the theft of more than $6.3 billion from Binance to North Korea's TraderTraitor hacking group. The agencies urged the crypto industry to block transactions linked to the stolen funds.

Hackers moved and swapped assets through thousands of automated transactions to obfuscate trails, primarily converting to Ethereum then using DeFi and swap services. About $3 billion has been frozen with industry help. Binance shut down temporarily, used reserves to cover losses, and later identified additional breach points beyond an initially reported compromised engineer laptop. A self-destruct mechanism hindered forensics.

Who is affected

Binance and its users; the broader cryptocurrency ecosystem including exchanges, DeFi platforms, and mixers that may process the stolen funds. Law enforcement and blockchain analytics firms tracking the laundering.

Why it matters

This is one of the largest crypto heists on record, demonstrating nation-state capability to target major exchanges and launder at high speed with automation. It pressures the entire industry on transaction monitoring, freezing cooperation, and operational security around high-value keys and infrastructure.

How it could have been prevented

Crypto platforms should immediately block listed TraderTraitor addresses and derived funds. Implement robust transaction monitoring for automated high-volume laundering patterns. Strengthen internal access controls, key management, and monitoring for engineer endpoints. Cooperate with law enforcement on freezes and share indicators.

What happened

Blockchain analytics firm Chainalysis detected a cybersecurity incident in which external attackers used "offensive agentic techniques" to access its systems and internal tools.

The company stated there was no indication of unauthorized access to investigation content, graphs, notes, cases, transaction results, clear-text passwords, MFA secrets, or session tokens. Attackers did obtain some customer user IDs, full names, email addresses, and organization role labels. The incident is unrelated to the recent Binance hack. Chainalysis is notifying affected customers and has engaged third-party experts.

Who is affected

Chainalysis customers, including crypto exchanges, banks, law enforcement, and government agencies whose user IDs, names, emails, and roles were accessed. The firm itself and its platform users face elevated social-engineering risk.

Why it matters

Exposure of customer contact and role data enables highly targeted phishing and social engineering against organizations involved in crypto crime investigations. The use of agentic techniques highlights how AI-driven attacks can target even specialized security and analytics providers.

How it could have been prevented

Affected customers should watch for unexpected password-reset or authentication prompts, enable MFA if not already active, and treat unsolicited contacts with caution. Chainalysis and similar firms should accelerate reviews of agentic attack surfaces, rotate any potentially exposed credentials, and harden internal tool access.

What happened

Google expanded its Cloud Vulnerability Reward Program with a new top-tier $250,000 reward for high-impact architectural research that demonstrates full administrative control over Google Cloud organizations or projects.

To qualify, the path must achieve roles/owner privileges or complete IAM control without relying on pre-existing project access, compromised credentials, or social engineering. RCE in Borg (Google's production infrastructure) or root-in-production findings are also eligible, with strict rules requiring researchers to stop immediately upon any Borg execution and forbidding further escalation.

Who is affected

Security researchers participating in Google's Cloud VRP. Indirectly benefits Google Cloud customers through improved isolation and architecture hardening from the research incentivized.

Why it matters

Higher bounties for tenant-isolation and cross-service logic flaws encourage deeper architectural scrutiny that traditional bug hunting may miss. Clear rules of engagement protect production while still attracting advanced research, including AI-assisted discovery.