Cybersecurity News

Microsoft has published a workaround for a known issue that blocks Windows 11 users from signing in with valid domain credentials after installing the September 2026 security updates. The problem affects domain-joined systems, where authentication failures occur despite credentials being correct. Microsoft's guidance offers a temporary mitigation while a permanent fix is developed, and the company has acknowledged the issue as known. Affected users should note: The issue appears after applying the September 2026 security updates on Windows 11. Login attempts with valid domain credentials can fail. Microsoft has provided a temporary workaround rather than a full patch at this time. Administrators managing domain-joined Windows 11 fleets are advised to review Microsoft's workaround and apply it where appropriate until a permanent resolution is released. Source: Unknown Has your organization encountered domain login failures since deploying the September 2026 updates, and are you applying Microsoft's workaround?

AI-Powered Attacks Are Reshaping Credential Theft — Identity Security Has to Keep Up Artificial intelligence is making credential theft faster and easier to scale, giving attackers far more opportunities to abuse valid identities. According to Specops, this shift means identity security can no longer stop at successful authentication — organizations must also verify that both the user and the device requesting access can be trusted. The core problem is that a correct username and password no longer proves much. When attackers use AI to accelerate and scale credential theft, compromised but legitimate-looking logins become one of the easiest paths into an environment. Traditional authentication checks may pass cleanly while the person — or machine — behind the request is not who they claim to be. This is why Specops argues for moving beyond authentication alone and toward validating trust on both sides of the access request: the identity and the endpoint it originates from. Key takeaways for defenders: Treat valid credentials as a starting point, not proof of trust. Verify the device as well as the user before granting access. Assume AI lowers the cost and time required to steal and reuse credentials at scale. Source: BleepingComputer Does your organization verify device trust alongside user credentials, or is authentication still the only gate?

The FBI has seized the domains behind NightmareStresser, one of the longest-running distributed denial-of-service (DDoS) for-hire platforms. The takedown targets a service that has been linked to thousands of DDoS attacks over its years of operation. NightmareStresser operated as a booter or stresser service, allowing customers to pay for attacks that flood targets with traffic in order to knock them offline. The platform has been connected to a large volume of DDoS activity, making it a long-standing fixture in the criminal DDoS ecosystem. Key points: The FBI carried out the domain seizure against NightmareStresser The platform is described as one of the world's longest-running DDoS-for-hire services It has been linked to thousands of DDoS attacks The operation removes the domains the service relied on to operate Details on charges, arrests, or additional infrastructure beyond the domain seizures were not provided in the source material. Source: Publication Name Has your organization ever been targeted by a booter or stresser service, and how do you defend against volumetric DDoS attacks?

Microsoft is investigating reports that the Windows 11 KB5124008 security update is breaking domain trust relationships on some enterprise systems, preventing users from logging in with valid domain credentials. According to the reports, affected machines lose the trust relationship between the workstation and its domain, which can block domain logins and require manual remediation to restore access. The issue appears to affect enterprise environments rather than typical home installations. Key details as they stand: Microsoft has confirmed it is investigating the reports. The affected update is KB5124008 for Windows 11. The impact is a broken domain trust relationship, blocking logins with valid domain credentials. No specific CVE ID has been provided in the source material, so none is included here. Administrators who suspect they are affected should review the current guidance from Microsoft and avoid broad deployment of the update until the investigation concludes. If domain trust is already broken, re-joining the affected device to the domain is the common recovery path, though it is worth checking official documentation before acting at scale. Note that Microsoft has not yet published a formal fix or full set of affected configurations, so treat any early workarounds with caution. Source: BleepingComputer Has your organization seen domain trust failures after deploying KB5124008, and if so, how are you handling it?

Researchers have disclosed a new hardware attack dubbed DDRop that defeats memory protection in Intel and AMD confidential computing environments. The technique works by silently dropping writes to a server's memory, causing the processor to keep reading stale encrypted data as though it were still current. According to the disclosure, carrying out the attack requires a specific starting position: the adversary must already control the server's software and be able to briefly access the machine in order to insert a small circuit. From there, the dropped writes undermine the integrity guarantees that confidential computing relies on. Key points: DDRop targets memory protection in Intel and AMD confidential computing. The mechanism involves silently dropping writes to server memory. The processor then continues reading old encrypted data as if it were up to date. Prerequisites: existing control over the server's software and brief physical access to insert a small circuit. Source: The Hacker News Does your organization run Intel TDX or AMD SEV-SNP workloads, and if so, what compensating controls are you considering while more details emerge?

U.S. cybersecurity and intelligence agencies have issued a joint accusation against China-based AI companies, alleging they are carrying out "systematic extraction" of proprietary capabilities from American frontier models through distillation attacks. The agencies describe the activity as occurring at an industrial scale and characterize it as forming the "core" of these firms' AI development strategies. Distillation, in this context, involves using the outputs of a more capable model to train a competing model, effectively transferring functionality without direct access to the original system's weights or training pipeline. The targeted models reportedly include Claude, GPT, Gemini, and Grok. Key points from the accusation: The activity is described as systematic rather than incidental, suggesting organized and sustained efforts. Agencies frame the extraction as central to the development approach of the accused firms, not a peripheral tactic. The targeting spans multiple major U.S. frontier model providers, indicating a broad rather than vendor-specific focus. The scale is characterized as industrial, implying automation and high-volume querying rather than manual probing. For defenders and security teams, this raises questions about API abuse detection, rate-limiting strategies, and monitoring for patterns consistent with large-scale model output harvesting. Organizations exposing LLM endpoints may want to review logging and anomaly detection around high-volume, structured query behavior. Source: The Hacker News Has your organization observed or implemented controls for detecting distillation-style API abuse against your AI services?

Wiz Research scanned internet-facing LiteLLM servers in February and found that almost 1 in 10 of them accepted sk-1234 — the example admin key printed in LiteLLM's own setup guide. LiteLLM is an open-source AI gateway, the layer a company places between its applications and the model providers it pays for. That key is the gateway's administrator credential, meaning anyone holding it can read every request passing through the gateway. Key points: Roughly 10% of the exposed LiteLLM instances Wiz Research scanned accepted the default example admin key sk-1234. The key appears as an example in LiteLLM's official setup documentation, so it is widely known and trivially guessable. Holding the admin key grants full access to the gateway's traffic, including the data flowing between internal applications and paid model providers. If your organization runs LiteLLM and it is reachable from the internet, verify that the admin key has been changed from the documented example and restrict exposure of the management interface. Source: The Hacker News Has your team audited its LiteLLM deployment for default or documented example credentials?

Over the past year, a new class of alert has been quietly climbing the queue in enterprise security operations centers: activity generated by AI tools and agents. These are not attacks targeting AI systems, but rather the everyday footprint of an organization simply using them. That footprint spans developers running coding agents through to non-technical staff signing consumer AI tools into corporate accounts, and it is growing faster than anything else in the alert stream. The challenge for SOC teams is that this traffic is not inherently malicious, yet it is unfamiliar and noisy. Traditional detection logic was never designed to distinguish routine AI-assisted work from suspicious behavior, which makes triage harder and increases the risk that genuine threats get lost in the volume. Alerts originate from legitimate AI usage, not from attacks against AI Contributors include developer coding agents and consumer AI tools tied to corporate accounts Growth rate outpaces other alert categories For defenders, the takeaway is that AI adoption across the business is now a security operations problem in its own right, and detection engineering needs to account for it. Source: The Hacker News Has your SOC started seeing alerts driven by internal AI tool usage, and how are you tuning detection to handle the noise?

Anthropic has disclosed that it identified and disrupted what it describes as industrial-scale illicit distillation attacks targeting its Claude model. According to the company, the activity originated from seven labs based in China, among them Alibaba, Moonshot, DeepSeek, Z.ai (also known as Zhipu), and MiniMax. It is worth noting that knowledge distillation is not inherently malicious. It is a legitimate machine learning training technique in which a large, capable AI model acts as a teacher to transfer knowledge to another model. The concern in this case stems from how the technique was allegedly applied and at what scale, which Anthropic characterizes as illicit. Key details as reported: Seven China-based labs were named in connection with the campaign. Named entities include Alibaba, Moonshot, DeepSeek, Z.ai (aka Zhipu), and MiniMax. Anthropic states it identified and disrupted the attacks against Claude. The activity is described as industrial-scale distillation. Knowledge distillation itself remains a legitimate training method; the dispute concerns its unauthorized use in this context. The report does not include specific CVE identifiers, advisory numbers, or indicators of compromise, so none are listed here. Organizations evaluating their own exposure to similar model-abuse campaigns should monitor vendor advisories and terms-of-service enforcement actions rather than rely on signature-based detection alone. Source: The Hacker News Do you think API-level rate limiting and output watermarking are enough to deter distillation attempts, or is stronger contractual and technical enforcement needed?

Identity was the primary target in roughly half of all confirmed malicious activity, according to Prophet Security, which investigated every alert across customer environments between May and July 2026. The analysis breaks down the four main attack patterns observed during that quarter and examines why some attacks succeeded while others were blocked. The findings highlight that attackers continue to concentrate on identity-based techniques, making credential theft, session abuse, and authentication manipulation central to modern intrusion attempts. The report also contrasts successful compromises against blocked attempts, offering insight into which defensive controls proved effective and where gaps remain. Identity was the target in approximately 50% of confirmed malicious activity Data covers customer environments from May to July 2026 Four main attack patterns were identified across the quarter The analysis explains why some attacks succeeded while others were blocked The research focuses on the difference between detection and prevention, noting that investigating every alert provided a clearer picture of attacker behavior than isolated incident reviews. Organizations relying on identity-centric defenses should treat these patterns as a reminder that authentication boundaries remain a primary battleground. The full breakdown of the four threat patterns and the reasons behind successful versus blocked attacks is available in the original report. Source: BleepingComputer How is your organization adapting its identity security controls in response to these findings?

Cybercriminals are increasingly hijacking AI user accounts through information stealer logs, turning stolen credentials into "stolen keys" that provide illicit access to tools from model providers such as Google, Anthropic, and others. Information stealers like Lumma Stealer and Vidar are built to harvest a broad range of data from compromised systems, including credentials, session tokens, and API keys. Because these tokens can be replayed, attackers may be able to bypass MFA protections and gain access to AI accounts without triggering typical authentication barriers. Key points: Attackers rely on infostealer logs to obtain replayable AI tokens. Lumma Stealer and Vidar are cited as examples of malware capable of harvesting credentials, session tokens, and API keys. Affected providers include Google, Anthropic, and other model providers. Replayable tokens can allow access that circumvents MFA. Source: The Hacker News Has your organization reviewed AI account tokens and session logs for signs of replay-based access?

The U.S. Department of Justice (DoJ) announced coordinated actions on Wednesday targeting an illicit online marketplace known as Xinbi Guarantee, which allegedly provided scam-as-a-service offerings to cybercriminals. The operation involved seizing Telegram channels used to administer the service and confiscating two cryptocurrency wallets tied to the scheme. Telegram channels linked to the marketplace’s operations have been taken down. Two digital wallets were frozen, containing approximately $52.8 million in cryptocurrency. The DoJ deployed the Scam Center Strike Force to Madagascar to assist in disrupting 13 scam compounds operated by Chinese organized crime groups in the region. The action highlights a growing focus on dismantling the infrastructure that enables large-scale fraud, rather than targeting individual perpetrators alone. The seized funds and shutdown of communication channels aim to cut off both the financial lifeline and operational coordination for these scam networks. Source: The Hacker News Has your organization encountered any scams linked to Xinbi Guarantee or similar marketplace-style fraud operations, and how are you approaching detection and prevention?

U.S. cybersecurity and intelligence agencies have accused six Chinese AI firms of running industrial-scale distillation attacks against American frontier AI models since at least late 2024. The operation reportedly siphoned billions of tokens from these systems — a scale that suggests coordinated, persistent targeting rather than isolated research experiments. Distillation attacks involve querying a model extensively to extract its outputs and behavior, then using that data to train a competing model at a fraction of the original cost. The agencies claim the Chinese companies leveraged this technique to effectively replicate the capabilities of leading U.S. models, potentially undermining the competitive edge of American AI developers. The extracted data could also be used to map model weaknesses or bypass safety guardrails. While the official statement does not name the specific models targeted, the scale of token extraction points to large-scale, automated abuse of API access. The advisory emphasizes that such attacks are not merely intellectual property concerns but also pose systemic risks to AI supply chains and national security. Affected organizations are urged to monitor for anomalous API usage patterns, such as high-volume requests from single sources, repeated identical prompts, or sudden spikes in token consumption. Monitor API logs for traffic originating from known hosting ranges or VPN endpoints. Implement rate limiting and anomaly detection on model endpoints to flag batch extraction attempts. Review access logs for patterns consistent with systematic prompt repetition. Enforce stricter authentication and usage policies for high-privilege API keys. The advisory reinforces a growing concern among security researchers that model distillation is evolving from a niche technical issue into a staple of state-sponsored economic espionage. This is not the first time U.S. authorities have warned about AI-related IP theft, but the explicit mention of billions of tokens marks a significant escalation in the reported scope of such operations. Source: Unknown Does your organization have visibility into API-level token usage, and what thresholds would you set to detect a distillation attempt before it reaches billions of requests?

Multi-factor authentication has raised the bar for direct account compromise, but attackers are adapting by targeting the processes designed to help users regain access when they are locked out. According to Specops, account recovery workflows—especially those handled through service desks—are becoming a preferred vector for social engineering because they often rely on weaker verification than the primary login flow. The issue is not the MFA itself, but the fallback mechanisms that bypass it. When a user calls the help desk claiming to have lost their phone or forgotten their password, the verification steps are frequently limited to personal questions, employee IDs, or other data that can be harvested from breaches or open-source intelligence. Once an attacker passes that checkpoint, they can reset the password and enroll their own authentication device, effectively taking over the account while the real user is locked out. Specops recommends treating the recovery process with the same rigor as the initial authentication. This includes verifying identity through multiple independent factors, checking the user’s location or device posture if possible, and requiring manager approval for high-privilege accounts. The service desk should also have clear procedures for detecting and rejecting requests that match known social engineering patterns, such as urgency, unfamiliar callback numbers, or inconsistencies in the user’s history. Key recommendations for hardening account recovery include: Requiring a secondary verification method that is independent of the one being reset (e.g., a hardware token or biometric check). Implementing time-based or context-based flags for recovery requests that occur outside normal working hours or from unexpected IP ranges. Establishing a mandatory waiting period or callback verification for password resets on admin or privileged accounts. Training service desk staff to recognize pressure tactics and to verify identity without relying solely on data that may be publicly available. The shift is notable because it exposes a gap in many security strategies: while MFA adoption reduces direct attacks, it can create a false sense of security if the recovery path remains weak. Attackers will continue to target the path of least resistance, and right now, that path often runs straight through the help desk. Source: Unknown How is your organization handling identity verification for service desk password resets—are you using more than just personal knowledge questions?

Microsoft has rolled out fresh cumulative updates for Windows 11, targeting both versions 25H2/24H2 and 23H2. The patches, identified as KB5124008 and KB5122880, are now being distributed to address a range of security vulnerabilities, correct existing bugs, and introduce new functionality to the operating system. For administrators and enthusiasts alike, the update cycle brings a mix of routine maintenance and feature enhancements. While the specific security flaws are not detailed in the advisory, the updates are marked as mandatory for keeping systems current against emerging threats. The new features vary by release channel and build version, so those on the Insider program may notice different additions compared to those on stable releases. If you are managing fleets of Windows 11 devices, keep an eye on your update logs for these two distinct KB entries. As with any cumulative update, it is advisable to back up critical data and test in a controlled environment before broad deployment, especially if your organization relies on specific hardware or third-party drivers that could conflict with new OS builds. Source: Unknown Are you seeing any unexpected issues or enhanced features after applying these updates on your test machines?

An exposed Advance Passenger Information System (APIS) database linked to Vietnamese aviation infrastructure left roughly 220 million passenger and crew records publicly accessible. The cache included names, passport numbers, dates of birth, nationalities, and flight details, with entries spanning 2017 to 2026. Researchers found the cloud-hosted system reachable via a public path that still accepted default credentials, allowing unrestricted read access without any authentication bypass or sophisticated exploit. The dataset reportedly covered both passengers and crew members, raising concerns regarding long-term identity theft and targeted phishing campaigns. Exposed data: full names, passport numbers, dates of birth, nationalities, and flight itineraries. Timeframe: records from 2017 through 2026. Root cause: default credentials left active on a cloud-based APIS instance. Access method: direct connection to the exposed system using vendor-default login details. The findings underscore persistent risks tied to misconfigured cloud deployments and unchanged factory settings, especially in border-control and travel infrastructure. It remains unclear whether the operator has restricted access or rotated credentials since the disclosure. Travelers whose data may be involved should monitor for unsolicited communications referencing flight history or passport details. Source: BleepingComputer For those in travel or aviation security, how is your organization auditing cloud-facing systems for default credential usage, and would a discovery like this prompt an immediate review of your APIS or PNR handling procedures?

Turning off images in your email client used to be a reliable way to block tracking pixels and malicious visuals. That safeguard is eroding, as attackers have found a workaround using QR codes constructed entirely from text characters. These codes render and remain scannable even when image loading is disabled, quietly undermining a precaution many users still depend on. Beyond that inbox annoyance, this week’s security landscape also featured a supply chain attack delivered through a trusted software repository, stealing credentials from unsuspecting developers. In the networking sphere, a protocol meant for secure management is being weaponized for router hijacks, and a fresh zero-day in Chrome demanded urgent attention from the patching community. Key developments from the past week: Google Chrome 0-day: A critical vulnerability was actively exploited in the wild. Details remain sparse, but the advisory points to a flaw in the browser's rendering engine that could lead to arbitrary code execution. Users are strongly advised to update to the latest stable build immediately. Router hijacking via TFTP: Attackers are abusing the Trivial File Transfer Protocol, a basic protocol often left enabled in embedded devices, to overwrite router firmware or configuration files. This allows for persistent control without triggering standard integrity checks. Coder-focused supply chain attack: Malicious code was injected into a popular package hosted on a well-known developer platform. The payload was designed to harvest environment variables and local credential stores, specifically targeting users with access to production CI/CD pipelines. For administrators and security teams, the response playbook remains consistent: prioritize patching for the browser flaw, audit network edge devices for exposed management interfaces, and review recent dependency updates for any anomalies. Source: The Hacker News Is your organization currently enforcing a policy that blocks or restricts TFTP traffic to network infrastructure, or is this something you are planning to address following this report?

Threat hunters have detailed a broad data theft and extortion campaign that is actively targeting Microsoft 365 and other software-as-a-service (SaaS) environments. The attack chain is notable for its reliance on help desk vishing, adversary-in-the-middle (AitM) token theft, and the use of residential-proxy infrastructure to mask malicious sign-ins. The campaign specifically singles out high-level corporate officers, including directors, vice presidents, and other executive staff. By impersonating legitimate IT support personnel, the attackers initiate phone calls to these high-value targets in an attempt to lower their guard and extract credentials or approve multi-factor authentication (MFA) prompts. Once initial access is established via stolen session tokens, the threat actors leverage residential proxies to make their traffic appear as if it originates from trusted, local IP addresses. This technique is designed to bypass geographic and risk-based conditional access policies that security teams often rely on. The ultimate goal of the operation is data exfiltration followed by extortion. The disclosure serves as a reminder that even with robust technical controls in place, social engineering remains a primary vector for compromising enterprise SaaS accounts. Primary Targets: Executive staff (Directors, VPs, C-suite). Initial Vector: IT help desk vishing (voice phishing). Key Technique: AitM token theft to bypass MFA. Evasion Method: Residential-proxy sign-ins to circumvent IP-based security policies. Impact: Data theft and subsequent extortion. Source: The Hacker News Given the focus on C-suite and executive roles, is your organization adjusting its help desk verification procedures or adding additional friction to MFA prompts for these specific high-risk users?

A phishing-as-a-service (PhaaS) operation tracked as BigBear 2.0 has been observed compromising Microsoft 365 environments at scale. According to telemetry, the campaign successfully bypassed multi-factor authentication (MFA) and harvested over 5,000 credentials across 258 organizations. The attack methodology relies on adversary-in-the-middle (AiTM) techniques to intercept session tokens. Below is a concise overview of the operational details and indicators. Target scope: 258 unique organizations; more than 5,000 Microsoft 365 accounts compromised. Primary tactic: AiTM phishing to steal session cookies, allowing attackers to bypass MFA requirements. Payload focus: Microsoft 365 login pages, often using lookalike domains and urgency-based lures. Post-exploitation: Stolen sessions are typically used for data exfiltration and follow-up business email compromise (BEC). Practical mitigation steps Enforce conditional access policies that require device compliance or trusted locations, not just a one-time password. Review sign-in logs for anomalous session activity, particularly where MFA was satisfied but geolocation or IP reputation appears suspicious. Implement phishing-resistant MFA, such as FIDO2 security keys, where feasible. Monitor for newly registered domains that mimic your organisation’s login portal. Given the low cost of entry for such PhaaS platforms, it is likely this pattern will continue to evolve. Organisations relying solely on OTP-based MFA should treat these campaigns as a direct threat to their identity perimeter. Source: Unknown Has your team reviewed sign-in logs for unusual session behaviour, or have you already moved to phishing-resistant MFA?

METR (Model Evaluation and Threat Research), a non-profit focused on assessing frontier AI models for long-horizon, agentic task performance, has disclosed two separate security incidents involving unauthorized external access attempts. The organization reports that no sensitive information is believed to have been compromised. In the more significant breach, attackers managed to steal a valid METR API key. This key was subsequently abused to consume AI credits worth approximately $600,000 before the threat was detected and mitigated. The stolen API key was used to run unauthorized model evaluations. METR has since revoked the compromised credentials and rotated related access tokens. The organization is reviewing its logging and monitoring to improve detection of similar abuse. The second incident did not involve the theft of credentials but is being treated as a notable attempt to probe METR's infrastructure. Both events have prompted a broader security review of their internal tooling and external integrations. While METR has not confirmed the specific vulnerability exploited, they emphasize that no proprietary research data or evaluation results were accessed. The financial impact is limited to the consumed compute credits, which are a direct cost of running large-scale AI models. Source: The Hacker News Is your organization tracking API key usage against abnormal spending spikes, or do you rely on static quotas that might delay detection of similar credential abuse?