CFToday Curated security signals.

Daily edition · 2026-08-27

Thursday, 27 August 2026

23 items across 5 sections, selected from 5969 candidates over 6 runs. 37 carried the panel unanimously.

Show
Section

  1. Srsly Risky Biz: China's AI-Enabled APT Operations Are Getting Interesting (opens in a new tab)

    Risky Business News ·fetched 27 Aug 2026, 07:38 UTC agreed2/2

    Why readArgues Chinese APT use of AI on malware development will degrade the code-similarity clustering that threat intel attribution depends on.

    Tom Uren and James Wilson discuss evidence that Chinese state groups are applying AI to malware development in a measured, practical way rather than for headline-grabbing autonomy, with the consequence that variant diversity makes clustering activity for attribution harder. They also cover the US disrupting Iranian operators by publicising, in a Treasury sanctions package, that some were hacking Iranian firms, and question whether burning that information in a sanctions document was the right call. Commentary rather than primary research, but the attribution point is worth carrying into intel workflows.

  2. Ransom & Dark Web Issues Week 4, August 2026 (opens in a new tab)

    AhnLab ASEC ·ATCP ·fetched 27 Aug 2026, 07:38 UTC agreed2/2

    Why readNames SAFEPAY ransomware hitting a South Korean industrial gas manufacturer and NoName057(16) plus BD Anonymous claiming DDoS against major Japanese organisations.

    AhnLab's weekly dark web roundup covers three items: user data from a Saudi Arabian streaming service offered for sale, a SAFEPAY ransomware attack on a South Korean industrial gas manufacturer and supplier, and DDoS claims against major Japanese organisations by NoName057(16) alongside BD Anonymous. The victim sectors and actor names are the usable content; IOCs and the detailed analysis sit behind an AhnLab TIP subscription. Regional tracking value for anyone with APAC exposure or watching hacktivist DDoS targeting.

  1. CISA Adds Six Exploited Flaws to KEV, Including NetScaler, Linux, and SQL Server Bugs (opens in a new tab)

    The Hacker News ·The Hacker News ·fetched 27 Aug 2026, 07:38 UTC Must read CVE-2019-1068 EPSS 44.7% agreed2/2

    Why readSix new KEV entries with a federal remediation deadline, including Citrix NetScaler ADC/Gateway CVE-2026-8452 and a 2019 SQL Server RCE that is still being exploited.

    CISA added CVE-2026-8452 (memory buffer flaw in NetScaler ADC and Gateway leading to denial of service), CVE-2019-1068 (Microsoft SQL Server RCE in the database engine service account context, EPSS 0.45), CVE-2022-0995 (Linux kernel out-of-bounds write for local privilege escalation), plus CVE-2015-5287 in Red Hat ABRT and CVE-2015-3246, on evidence of active exploitation. The age of several entries is the signal: unpatched Linux and Red Hat hosts from a decade ago are being used for local escalation today. KEV membership means confirmed in-the-wild use and a binding remediation clock for federal agencies.

  2. CVE-2026-76835 (CVSS 9.3): OAuth2 Proxy honours a client-supplied X-Forwarded-Uri header when deciding whether a request may skip authentication, because the guard added for CVE (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC Must read CVE-2026-76835 CVSS 9.3 EPSS 0.3% agreed2/2

    Why readOAuth2 Proxy trusts a client-supplied X-Forwarded-Uri when deciding whether to skip authentication, and in default reverse-proxy mode every client counts as a trusted proxy, so skip_auth_routes becomes an unauthenticated bypass of any protected upstream path.

    GetRequestURI in pkg/requests/util/util.go prefers the X-Forwarded-Uri header over the real request URI whenever CanTrustForwardedHeaders returns true, and isAllowedPath matches skip_auth_routes and skip_auth_regex against that spoofable value. buildTrustedProxyNetSet in pkg/apis/middleware/scope.go falls back to defaultTrustedProxyIPs of 0.0.0.0/0 and ::/0 when reverse proxy mode is enabled without trusted_proxy_ip set, which makes the guard added for CVE-2026-40575 inert in the default configuration. An unauthenticated attacker requests a protected path with X-Forwarded-Uri set to an allow-listed route, the skip-auth decision is made on the spoofed path, and the upstream receives the real one; anyone fronting internal apps with oauth2-proxy in Kubernetes should set trusted_proxy_ip now and audit their skip-auth lists.

  3. CVE-2026-71921 (CVSS 9.3): Multiple DrayTek VigorSwitch models contain a pre-authentication command injection vulnerability in the setget.cgi interface. The vulnerability is cau (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 03:36 UTC Must read CVE-2026-71921 CVSS 9.3 EPSS 3.2% agreed2/2

    Why readPre-authentication command injection giving root on DrayTek VigorSwitch models through the pass field of setget.cgi, no credentials needed.

    CVE-2026-71921 (CVSS 9.3, CVSS:4.0/AV:N/AC:L/PR:N/UI:N) affects multiple DrayTek VigorSwitch models: setget.cgi fails to filter the pass field before passing it to command execution, so a remote unauthenticated attacker gets arbitrary commands as root. EPSS is only 0.032 but sits in the 87th percentile, and DrayTek network gear has a long history of mass exploitation once a pre-auth path is public. Treat any internet-reachable or management-plane-exposed VigorSwitch as urgent and confirm the management interface is not routable.

  4. CVE-2026-39975 (CVSS 9.4): Combodo iTop is a web-based IT service management tool. Prior to 3.2.3, unauthenticated users could delete the .readonly file on iTop instances, leadi (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC CVE-2026-39975 CVSS 9.4 EPSS 0.3% agreed2/2

    Why readIn Combodo iTop before 3.2.3, deleting the setup-created .readonly file removes the write guard and opens a path to code execution.

    iTop prior to 3.2.3 lets unauthenticated users delete the .readonly file that the setup process creates to block write actions, and removing it leads to code execution on the instance. CVSS 4.0 puts it at 9.4 with high confidentiality, integrity and availability impact and subsequent-system impact set high; EPSS is 0.0035. iTop instances are frequently internet-facing service desks holding CMDB and credential data, so 3.2.3 is the upgrade and the presence of .readonly is a quick post-patch check.

  5. CVE-2026-76838 (CVSS 8.4): Hi.Events validates a webhook destination only when it is registered, never when it is used. NoInternalUrlRule in backend/app/Validators/Rules/NoInter (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC Research CVE-2026-76838 CVSS 8.4 EPSS 0.3% agreed2/2

    Why readA validate-on-registration, never-on-use webhook design in Hi.Events yields full-read SSRF to cloud metadata via redirects and DNS changes.

    NoInternalUrlRule in backend/app/Validators/Rules/NoInternalUrlRule.php resolves the hostname with gethostbyname() and blocks private ranges only at registration time; WebhookDispatchService then calls the stored URL through spatie/laravel-webhook-server with no Guzzle options set, so redirect following stays on and the resolution is never repeated. A destination that redirects to loopback, RFC1918 or the cloud metadata endpoint, or whose DNS record simply changes later, gets fetched by the server, and WebhookResponseHandlerService stores the response body where the requester can read it back from the webhook logs endpoint. The pattern generalises well beyond this product: SSRF filters that run at configuration time and not at dispatch time are not filters.

  6. CVE-2026-71504 (CVSS 8.6): Dolibarr before 24.0.0 contains an improper authorization vulnerability in the Members REST API that allows attackers with only member-creation rights (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC CVE-2026-71504 CVSS 8.6 EPSS 0.3% agreed2/2

    Why readAny Dolibarr user with member-creation rights can overwrite the administrator's password through the Members REST API.

    Dolibarr before 24.0.0 never checks password-change permission in the Members REST API, so a low-privileged account can supply an arbitrary user id and new password in the request body and take over the system administrator account, locking the real owner out. Full admin compromise from a minor role in an ERP that many small organisations expose to the internet. Fix is upgrading to 24.0.0; EPSS is currently 0.003.

  7. CVE-2026-9254 (CVSS 8.7): An unauthenticated OS command injection vulnerability exists in the parental control functionality of Archer BE800 V1, BE3600 V1, and AX75 V1 due to i (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC Must read CVE-2026-9254 CVSS 8.7 EPSS 2.3% agreed2/2

    Why readUnauthenticated OS command injection in the parental control feature of TP-Link Archer BE800, BE3600 and AX75, giving a LAN attacker root on the router.

    Improper filtering of special characters in parental control parameters on Archer BE800 V1, BE3600 V1 and AX75 V1 lets an attacker on the local network inject commands that execute as root, with no authentication needed. CVSS 4.0 scores it 8.7 (AV:A) with high impact on the device and low subsequent impact on the network; EPSS is 0.023, the 82nd percentile. These are high-volume consumer and small-office units, and root on the gateway means traffic interception, so a compromised laptop or guest device on the LAN is enough to pivot to persistent network control.

  8. ZDI-26-602: Foxit PDF Reader Doc Object Use-After-Free Remote Code Execution Vulnerability (opens in a new tab)

    ZDI Published Advisories ·fetched 27 Aug 2026, 07:38 UTC CVE-2026-13128 EPSS 0.2% agreed2/2

    Why readUse-after-free in Foxit PDF Reader Doc object handling gives remote code execution in the reader process; CVE-2026-13128 is patched.

    ZDI-26-602 describes a use-after-free in Foxit PDF Reader arising from failure to validate that a Doc object exists before operating on it, allowing code execution in the context of the current process. Exploitation requires user interaction: the target must open a malicious file or visit a malicious page. Foxit has shipped an update, and EPSS of 0.0017 suggests no current exploitation, so this is normal patch-cycle work for estates that deploy Foxit rather than Adobe.

  9. CVE-2026-71918 (CVSS 8.6): Multiple DrayTek VigorSwitch models contain a command injection vulnerability in the webBackupAction function. The vulnerability is caused by insuffic (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 03:36 UTC CVE-2026-71918 CVSS 8.6 EPSS 3.1% agreed2/2

    Why readPost-authentication root command injection in DrayTek VigorSwitch webBackupAction, one of a cluster disclosed alongside a pre-auth bug in the same web interface.

    CVE-2026-71918 (CVSS 8.6, PR:H) affects multiple DrayTek VigorSwitch models: the webBackupAction function fails to filter the option, key, pw_encode, pathN and valueN fields before command execution, giving arbitrary commands as root to anyone with web admin credentials. On its own this is an admin-to-root escalation of limited interest, but it chains cleanly behind the pre-auth setget.cgi injection (CVE-2026-71921) in the same disclosure set. Patch the whole cluster together.

  10. CVE-2026-71933 (CVSS 8.8): Multiple DrayTek VigorSwitch models contain unauthorized operation vulnerabilities in multiple syslog functions. The vulnerability is caused by missin (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 03:36 UTC CVE-2026-71933 CVSS 8.8 EPSS 0.4% agreed2/2

    Why readUnauthenticated attackers can clear logs, restart services and rewrite startup configuration on DrayTek VigorSwitch through missing authorization checks on syslog functions.

    Several syslog-related functions across multiple VigorSwitch models perform no authorization check, so crafted requests from an unauthenticated remote attacker can modify configuration, restart services, write startup config and clear logs. Log clearing without credentials is the part worth flagging: it removes the evidence trail on the switch itself before anything else is investigated. This is the only pre-auth write primitive in the DrayTek batch (CVSS 4.0 vector shows PR:N, VI:H/VA:H), though EPSS remains low at 0.0038.

  11. CVE-2026-71919 (CVSS 8.6): Multiple DrayTek VigorSwitch models contain a command injection vulnerability in the sysreboot function. The vulnerability is caused by insufficient f (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 03:36 UTC CVE-2026-71919 CVSS 8.6 EPSS 3.1% agreed2/2

    Why readAnother authenticated root command injection on DrayTek VigorSwitch, this one via the config, act, pathN and valueN fields of the sysreboot function.

    CVE-2026-71919 (CVSS 8.6, PR:H) is a command injection in the sysreboot function of multiple DrayTek VigorSwitch models, caused by insufficient filtering of the config, act, pathN and valueN parameters. It executes as root but requires valid web management credentials. It is part of the same DrayTek disclosure batch as the unauthenticated setget.cgi bug and should be remediated in the same firmware update.

  12. CVE-2026-71924 (CVSS 8.6): Multiple DrayTek VigorSwitch models contain a command injection vulnerability in the getVid function. The vulnerability is caused by insufficient filt (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 03:36 UTC CVE-2026-71924 CVSS 8.6 EPSS 3.1% agreed2/2

    Why readDrayTek VigorSwitch getVid passes username and password straight into command execution, yielding root for an authenticated attacker.

    CVE-2026-71924 (CVSS 8.6, PR:H) affects multiple DrayTek VigorSwitch models: the getVid function does not filter the username and password fields before command execution, so crafted credentials produce arbitrary command execution as root. Administrative web credentials are required. The pattern of injecting through login-shaped fields recurs across several CVEs in this batch, which suggests a shared unsanitised helper rather than isolated mistakes.

  1. From Fleet to Lab: Revisiting the Security and Complexity of Industrial Rowhammer Mitigation (opens in a new tab)

    arXiv cs.CR (all) ·Hritvik Taneja, Moinuddin Qureshi ·fetched 27 Aug 2026, 03:36 UTC Must read Research agreed2/2

    Why readBreaks Sigries, the memory-controller Rowhammer defense Microsoft shipped in the Azure Cobalt 200 SoC, with a sub-bank Round-Robin Attack that cuts mean time to failure to about one second.

    Sigries pairs an under-provisioned Misra-Gries tracker with a row-sampling fallback and assumed the sampling-to-tracker transition was always safe. A Round-Robin Attack spread across sub-banks exploits that transition and drops MTTF to roughly 1 second, eight orders of magnitude below the 13 years PARA achieves, alongside the CAM complexity and storage overhead Sigries already carries. The authors propose FiRM, which filters before tracking so the tradeoff between tracking storage and mitigation rate no longer forces an insecure fallback.

  1. Breaking Claude Code Opus 5 Auto Mode (opens in a new tab)

    Embrace The Red ·fetched 27 Aug 2026, 07:38 UTC Must read Research agreed2/2

    Why readA website summary request hijacks Claude Code Opus 5 in Auto Mode to code execution at a 60-80% success rate, against a commissioned evaluation that reported 0.00%.

    Auto Mode, which since mid-August is the default starting mode in Claude Code, replaces human approval prompts with a safety classifier. The author shows that indirect prompt injection delivered through a benign-looking summarise-this-page request drives code execution in 60-80% of trials on a small sample, directly contradicting a third-party evaluation commissioned by Anthropic that reported a 0.00% prompt injection success rate for the same configuration. The gap between vendor-commissioned assurance numbers and adversarial testing is the finding that matters for anyone running coding agents with reduced approval friction.

  2. CVE-2026-76072 (CVSS 8.3): The Continue CLI applies an incomplete denylist as its only barrier to destructive shell commands when running unattended. In headless mode and auto m (opens in a new tab)

    NVD ·fetched 27 Aug 2026, 07:38 UTC Must read Research CVE-2026-76072 CVSS 8.3 EPSS 0.3% agreed2/2

    Why readShows exactly how a denylist guardrail on an agentic coding CLI fails: rm -rf /home, /var, /opt and even rm -rf $HOME all pass Continue's terminal-security check in headless and auto mode.

    In Continue CLI's headless and auto modes, defaultPolicies.ts grants Bash the allow permission and permissionChecker.ts only hard-blocks on a disabled verdict, leaving isCriticalCommand in packages/terminal-security as the single control. Its dangerous-path list covers only /, ~, /usr, /etc, /bin and /sbin, so recursive deletion of /home, /root, /var, /opt or /srv passes, and because the command is parsed with shell-quote, $HOME collapses to an empty token before the shell re-expands it at spawn time. find -delete is rated high risk rather than blocked, and shred, wipefs, truncate and pkexec are unhandled, which matters because the agent ingests untrusted web pages and repository content. Worth reading by anyone running any coding agent unattended: the shell-quote expansion gap is a general class of bypass, not a Continue-specific bug.

  3. Vulnerable Code Search: Transferable Attack for Code Language Models (opens in a new tab)

    arXiv cs.CR (AI) ·Kaicheng Wang, Liyan Huang, Jesse Thomason, Weihang Wang ·fetched 27 Aug 2026, 07:38 UTC Must read Research agreed2/2

    Why readA functionality-preserving identifier rewrite makes arbitrary code rank as a match for a target query, cutting MRR of state-of-the-art code retrieval by up to 77%.

    The attack perturbs only identifiers in a snippet, leaving behaviour unchanged, to artificially align it with a chosen search query. Adversarial examples computed against a small open model (CodeT5+) transfer to closed embedding models such as Voyage-code-3 and to Gemini-3.1-Pro, so an attacker does not need access to the target retriever. The practical consequence is a poisoning path into AI code search and RAG-backed coding assistants: plant a snippet that surfaces for common developer queries and it gets reused.

  4. RTLGuard: A Lightweight Teacher-Student Defense for Poisoned RTL Code Generation Models (opens in a new tab)

    arXiv cs.CR (AI) ·Mahshid Rezakhani, Kimia Azar, Hadi Kamali ·fetched 27 Aug 2026, 03:36 UTC Research agreed2/2

    Why readA teacher-student sanitisation method for removing backdoors from fine-tuned RTL code generation models without full retraining.

    RTLGuard addresses backdoored LLMs that emit hardware Trojans when triggered by benign-looking prompts, a risk introduced when designers use opaque third-party fine-tuned models. Instead of full-parameter retraining, it fine-tunes a small clean teacher on trusted RTL data and steers the poisoned student with a composite objective. Narrow to silicon design teams, but a concrete defence for model supply-chain poisoning where retraining cost is prohibitive.

  1. Boston Scientific hit by cyberattack causing ‘global disruption' (opens in a new tab)

    Google News: incidents · NBC Boston ·fetched 27 Aug 2026, 07:38 UTC Must read agreed2/2

    Why readBoston Scientific, a major medical device manufacturer, confirms a cyberattack causing global operational disruption, the kind of event peers in medtech and healthcare supply chains will be asked about this week.

    NBC Boston reports the attack is disrupting Boston Scientific operations globally. No attribution, malware detail or timeline has been published yet, so the fact of a confirmed multi-site disruption at a named device maker is the story. Expect questions about device supply continuity, supplier notification obligations and whether any regulatory filing follows.

    Also covered byCBS News (opens in a new tab),Fierce Biotech (opens in a new tab),MedCity News (opens in a new tab).

  2. Anthropic tells staff to work from home due to possible security team strike (opens in a new tab)

    Hacker News ·DGAP ·fetched 27 Aug 2026, 03:36 UTC 123 points agreed2/2

    Why readAnthropic told staff to work from home over a possible strike by its security team, a named-company operational security disruption a board will ask about.

    Business Insider reports Anthropic instructed employees to work remotely because of a potential labour action by its security staff. The item is physical and personnel security rather than a technical incident, but it is a concrete event at a named organisation with a date, and it sits in the awkward space between labour relations and site security that executives are rarely prepared to answer for. Feed text was thin, so treat detail beyond the headline claim as unconfirmed.

  3. Some families have faced funeral delays after cyberattack on TruStage insurance (opens in a new tab)

    Google News: incidents · First Alert 4 ·fetched 27 Aug 2026, 03:36 UTC agreed2/2

    Why readA cyberattack on TruStage, a major credit-union insurance provider, is delaying funeral payouts, which is the kind of downstream consumer harm that turns an IT incident into a board and press problem.

    TruStage is reporting service disruption from a cyberattack severe enough that families waiting on life insurance and final-expense payouts have had funerals delayed. The report carries no technical detail on the intrusion. For insurers and their vendors, the story is the operational blast radius: claims processing outages surface publicly through affected customers long before any breach notification does.

  4. Government admits water system cyberattacks were worse than first reported (opens in a new tab)

    Google News: incidents · NBC 5 Chicago ·fetched 27 Aug 2026, 03:36 UTC agreed2/2

    Why readGovernment now says the scale of cyberattacks on US water systems exceeded what was originally disclosed, which resets the baseline for anyone briefing on critical-infrastructure exposure.

    Officials have acknowledged that intrusions into water utility systems were more extensive than earlier public accounts indicated. The report is thin on which utilities or which actors are involved. It matters chiefly as a revision to the official record that water sector operators and their regulators will be asked about.

Unverified claims posted by extortion groups on their own leak sites, not confirmed breaches. Listing is the group's assertion; many named organisations have not disclosed an incident, and some entries are false or recycled. Leak sites are never linked from here.

Claimed victimGroup SectorCountry Seen
ETNA Software Eclipse Technology IT 27 Aug 2026
singleton.com chaos - AU 27 Aug 2026
Ne...n M... SilentRansomGroup - - 27 Aug 2026
Globalport Terminals qilin Transportation PH 27 Aug 2026
Agrimac Storm Agriculture and Food Production AU 27 Aug 2026
Sprachakademie Rhein-Ruhr Storm Education DE 27 Aug 2026
Otto Sieve GmbH Storm Manufacturing DE 27 Aug 2026
ITD Informations technologie Storm Technology DE 27 Aug 2026
Our Hospice Of South Central Indiana Storm Healthcare US 27 Aug 2026
National Salvage Storm Manufacturing US 27 Aug 2026
W****s payoutsking - US 27 Aug 2026
Zion Construction thegentlemen Manufacturing US 27 Aug 2026
BENCIVIL incransom Professional Services US 27 Aug 2026
fpmanagement.nl lockbit5 Professional Services NL 27 Aug 2026
G... T... SilentRansomGroup - - 27 Aug 2026
Kling Automaten qilin Retail & E-Commerce DE 27 Aug 2026
Dotlines qilin Technology SG 27 Aug 2026
Ipro.com(revealdata.com) customer DB + full database backup emperador Technology - 27 Aug 2026
Rohloff Group incransom Manufacturing ZA 27 Aug 2026
Cetylite akira Manufacturing US 27 Aug 2026
CGP MEP akira - - 27 Aug 2026
Seabrook Island akira Hospitality US 27 Aug 2026
Ruby Seven Studios incransom Technology US 27 Aug 2026
GPS Grothkopp und Partner qilin Professional Services DE 27 Aug 2026
theheartcenterofmemphis.com lockbit5 Healthcare US 27 Aug 2026
How this edition was made
Candidates fetched
5969
New after deduplication
720
Kept by the panel
37
Published
37
Generated
27 Aug 2026, 23:36 UTC