Full-text search across article titles and summaries. Refine with sector, year, content type, and reading-time filters.
CISA has published its first-ever guidance on setting up honeypot systems, accounts, and data to distract potential attackers and ultimately help discover them. The guidance comes from internal discussions with CISA threat hunters and pentesters who found that cyber decoys can be a low-cost, high-fidelity way to detect attackers who have gained access to a network.
A New Jersey court transferred 14 Radaris-linked domains to Atlas Data Privacy after defendants missed court deadlines in Daniel's Law litigation.
CVE-2026-89026 lets unauthenticated attackers forge JWT bearer tokens in Issabel Framework.
Prominent Rust maintainers are being lured onto fake job and contract video calls that end with a request to install a missing audio codec or paste a clipboard command.
A human attacker exploited CVE-2026-39987 to achieve pre-authentication remote code execution and harvest cloud credentials from Marimo notebooks. The Sysdig Threat Research Team observed the operator pivot from the vulnerable terminal WebSocket endpoint to an SSH bastion host in eight seconds to access an AWS Secrets Manager SSH key without using large language models. Defenders must focus detection efforts on the shape of the attack chain rather than specific operator fingerprints, because prompt injection traps designed for AI agents failed to identify the human actor.
China-linked UNC3569 used a single crafted link to gain system-level code execution through Sogou Input Method, Tencent's Chinese-language keyboard app used by hundreds of millions of Windows users, and deployed its GrayRabbit backdoor. Gen Threat Labs found that CVE-2026-51990 chained argument injection in the sgbiz protocol handler with a bundled Chromium 80 browser that was six years out of date, ran without a sandbox, and dropped same-origin protections. Tencent patched the injection bug in version 16.3.0.3498, but the outdated Chromium engine behind the chain remained unchanged as of September 10.
CenterPoint Energy confirmed an attacker pulled customer data through an external-facing API on September 12. A hacker using the alias "4d722e4d656f77" claims 7.49 million records, including names, phone numbers, billing addresses, account numbers, payment status, and partial Social Security numbers. The attacker says the API lacked WAF protection, rate limiting, and JWT/auth tokens, and stopped only after a CAPTCHA was added.
JA4 is a TLS fingerprint that shows how a client connects, not just where it connects to. Unlike JA3, part of it is readable, so hunters can see the TLS version, SNI, cipher count, and ALPN at a glance. The post includes KQL queries to pull these values from Entra ID sign-ins and Defender network logs, so teams can learn what normal looks like and spot clients that don't match.
A £3 Z05L WiFi extender uses an MT7620 chip and a four-megabyte Winbond flash. Its firmware starts Telnet on every boot with root password 198277tt@, and the web API accepts requests without session checks, so a mac parameter can reach system() without validation, allowing root command injection. Firmware updates verify CRC32 values only, allowing forged images, and the firewall leaves inbound services exposed on bridged networks as the config partition stores WiFi credentials in plaintext.
CVE-2026-90894 lets a non-admin macOS user gain root through Parallels Desktop's appliance installer. JFrog tested Desktop 26.4.0 build 57513 on Apple Silicon. The root daemon exposes a world-writable Unix socket where an unsigned local process can authenticate through PrlSrv_LoginLocal. A quote in sVmParentPath injects tar's --use-compress-program option, and tar then launches an attacker script as root. Parallels released version 27.0.0 on September 1 to fix this issue.
Sponsored
If you're reading this ad, you know they work. TLDR subscribers make a deliberate choice to open their inbox every weekday. You reach security engineers, CISOs, and infosec leaders at the precise moment they're most receptive to new information shaping their work. Become a TLDR sponsor. [10]
Entrust added Cryptographic Bill of Materials (CBOM) import and export to its Cryptographic Security Platform, so teams can map which systems depend on weak or noncompliant keys and certificates and fix the riskiest first. The update also brings Ansible-based certificate automation, composite algorithms for phased post-quantum migration, and SPIRE support for AI agent and workload identities. The platform now runs as a service or on-premises, which matters as DORA, NIS2, and the recent US Executive Order push for CBOM-based inventories.
Hackuity consolidates vulnerability and asset data from 130+ security products and ranks flaws by severity, exploitability, asset criticality, threat activity, and business impact. Teams can assign and coordinate remediation across security, IT, cloud, and engineering.
vol-rs is a port of the Volatility 3 memory forensics framework from Python to Rust.
WSO2 API Manager CVE-2026-5430 is under active attack. Forged JWTs using unsupported signing algorithms can bypass authentication and grant administrator access in affected deployments, including API Manager 4.1.0–4.6.0, API Control Plane, Traffic Manager, and Universal Gateway. Captured tokens carried admin privileges, and attackers may access backend endpoints, credentials, consumer keys, application secrets, and internal services.
An in-depth guide to how the author decided to reverse-engineer Apple's Find My system to register a Linux device and (consensually) track a friend. The author first started by attempting to use the iCloud web API but quickly discovered that it didn't support Find My. After reversing and fuzzing the flow, the author was able to reproduce the process by: authenticating to Apple's private service, registering the Linux machine as an IDS client, receiving the existing Find My share key, and using it to fetch and decrypt the friend's latest location.
A new AI regulatory framework, the FRONTIER Act, includes a provision that would require third parties to audit the safety of new frontier models. Rep. Josh Gottheimer pushed back and stated that frontier models should instead need to be submitted to the federal government for a short-clocked review cycle before release.
Oracle's September 2026 patch update fixed over 800 vulnerabilities across 17 product families, including more than 100 critical flaws and over 240 exploitable remotely without authentication.
Attackers are exploiting CVE-2026-76461, a 9.8 flaw in Cisco Secure Email Gateway appliances.
Logi Options+, Logitech's widely used app for configuring its mice and keyboards, allowed any standard Windows user to reach SYSTEM because its privileged updater trusted only pipe connections from an agent process the user could inject. Researchers at AmberWolf used that access to override the {depot} path via the reinstall IPC endpoint, which directed a SYSTEM msiexec call to an unsigned MSI that skipped signature checks because the manifest set isSystem to true. The bug is tracked as CVE-2026-12518 and was patched on August 19, so admins should confirm that fleets run a newer build and alert on msiexec spawned by logioptionsplus_updater.exe from user-writable paths.
Attackers hijacked HBO Max's Reddit advertising account and posted malicious ads. The ads led to fake HBO Max pages with CAPTCHA-style ClickFix prompts that made the victims copy commands into Windows Command Prompt, PowerShell, or macOS Terminal. The commands installed infostealers that targeted passwords, active sessions, and crypto wallets. Reddit paused the ads three days after a user flagged them, though how the account was taken over remains unknown.
Attackers scanned exposed Vite development servers in August and exploited CVE-2026-39364. Crafted query parameters bypassed server.fs.deny and returned protected files through /@fs/. Observed targets included .env files, AWS credentials, Azure profiles, Terraform state, database passwords, and /proc data with requests spoofing crawler user agents and forged forwarding headers.
FulcrumSec, a financially motivated data extortion gang behind breaches at Manchester Airports Group, Arup, and Novo Nordisk, told Lab539 it runs up to 250 servers scanning client-side JavaScript and fresh bugs like React2Shell, then bulk-tests harvested credentials before picking victims. From there, it chains pivots through GitHub repos, AWS, Azure, and Secrets Manager without touching an endpoint, with a single Arup PAT reportedly opening about 10,000 private repositories. Defenders should baseline normal identity behavior and flag tokens that suddenly clone thousands of repos or enumerate unrelated secrets from unfamiliar infrastructure, since this activity relies on valid credentials rather than malware.
A malicious workspace can place a core.fsmonitor hook in .git/config. Cursor CLI ran internal Git commands outside its macOS Seatbelt sandbox. A read-only prompt triggered git ls-files, which executed the hook. The proof payload wrote to $HOME, while sandboxed shell commands could not. Anysphere acknowledged the report and shipped Cursor CLI 2026.08.04-aaa8809, and the fix injects GIT_CONFIG environment settings into every Git spawn, disabling core.fsmonitor, custom hooks, and repository attributes.
Unit 42 clustered the CloudTrail activity of over 40,000 identities across 125 AWS environments and found they fall into consistent functional roles like console admins, CI/CD runners, backup agents, and security tools, regardless of naming or assigned IAM policies. That behavioral baseline matters because attackers masquerade behind benign labels and existing permission profiles, so a backup service suddenly running admin actions stands out even when its permissions allow it. The approach distills into a lightweight classifier that runs as standard SQL, letting teams tag identity roles at scale without a full ML pipeline.
Sponsored
OpenHunterAI is a local AI red-team workspace that runs attacker-style assessments against web apps, APIs, and LLM applications you own, combining browser inspection, recon, ZAP, and a Nuclei adapter under one scan plan. Its differentiator is scope control, with domain verification, deterministic plans, and human approval gates before any sensitive validation runs. It also ships as a skill for Claude Code, Codex, and Gemini CLI, so coding agents can prepare scoped assessments that wait for human reviewers to sign off.
A powerful multi-platform RF toolbox that deploys specialized radio, hardware, and other security tools in seconds on Linux, Windows, and macOS - supporting x86_64, ARM64 (Raspberry Pi and Apple Silicon), and RISC-V architectures without disrupting your primary OS.
AIUC audits and certifies deployed agents against its AIUC-1 standard. Its service runs roughly 5,000 tests covering jailbreaks, hallucinations, and data leaks. Customers receive a 100-page report, with humans validating final audit results.
Roei Sherman evaluated 33 local models on a two-node DGX Spark cluster for red teaming capabilities. The evaluation consisted of a security capability that tested whether the models can reason by testing them with a three-phase red team scenario, a red team benchmark to test the willingness and accuracy of model responses, and a coding benchmark consisting of Aider Polyglot plus two custom suites. Sherman found that MoE models consistently performed above their model class across all three benchmarks, with Qwen3.6-35B-A3B (FP8) winning the security capability benchmark, Qwen3-Coder-Next (FP8) winning the coding benchmark, and more mixed results on the red team benchmark.
Mythic, an open-source C2 framework that lets operators mix agents, transport profiles, and a shared web UI, is popular with red teams but has also turned up in real intrusions, according to new Censys research. Censys tracked 131 internet-exposed Mythic hosts, and 115 of them still carried the default O="Mythic" or O="Mythic C2" certificate. One cluster Censys flagged paired Mythic with custom Rust implants, Discord-based C2, and a domain disguised as Windows telemetry traffic.
Sponsored
33% of developers using AI agents say their organization has experienced a breach tied to overprivileged non-human identitiesreport security incidents from overprivileged NHIs. Learn why traditional
Apple patched more than 200 macOS vulnerabilities across Golden Gate 27, Tahoe 26.7, and Sequoia 15.8, including kernel code execution bugs in AVEVideoEncoder and UDF, as well as remote code execution paths via Bluetooth, CUPS, and WebDAV.
Japan's Digital Agency found that attackers exploited a publicly disclosed VPN flaw.
WordPress.org scans each plugin release during a six-hour cooldown before update distribution.
China's National Computer Virus Emergency Response Center is warning of "skill poisoning" attacks, where counterfeit plugins compromise AI agents to secretly download malware. Analysts have already identified eight malicious skill packs in popular agent repositories that impersonate legitimate tools, granting attackers unauthorized file access and remote-control capabilities. These poisoned plugins exploit an AI agent's natural task-execution design, bypassing the need for a user to actually click a suspicious link.
Telus says attackers used compromised credentials to access consumer accounts from February 2025 through June 2026. Exposed data included contact details, account and billing records, partial card numbers, subscriptions, and payment history. Attackers attempted service transfers and changed some services. Telus reset credentials, added monitoring, notified police, and offered identity-theft protection.
Attackers are exploiting CVE-2026-42018, CVE-2026-42016, and CVE-2026-82329 in JFrog Artifactory. Activity includes admin-account creation, Rust backdoors, Groovy plugins, web shells, token minting, and key theft. Wiz found 59 percent exposed to CVE-2026-42016 and 62 percent to CVE-2026-42018, so please patch exposed instances, restrict network access, and audit privileged logins.
Generic security prompts and OWASP-style checklists failed to improve AI coding security and occasionally increased the frequency of high-severity bugs when the Monad Foundation engineering team tasked an assistant with writing secure code for a URL unfurler. Utilizing an expert prompt that incorporates a specific application threat model reduced validated security findings by approximately 43 percent. Defenders must replace generic checklists with expert prompts incorporating specific application threat models to harden AI-generated code.
Anthropic's threat intelligence report demonstrates that artificial intelligence lowers the cost of familiar cyber attacks against targeted organizations. The technology automates reconnaissance, retooling loops, and campaign coordination. Defenders must reassess their controls against repeated, parallel, and fast-adapting attack attempts despite the continued expense of compute resources and specialized exploits.
Socket found that a cross-store browser extension, 'Twitch Enhanced Viewer | JeetBot', captures account OAuth tokens from Twitch pages. Current Chrome and Firefox builds add tokens to proxy redirect URLs for most viewed channels, but earlier builds sent tokens directly to set-token endpoints. The extension has about 31,000 installs across both stores, and anyone holding a token can access chat, whispers, account settings, and channel points. Remove the extension, disconnect Twitch sessions, and re-authenticate.
Sponsored
Traffic skyrocketing? If you're on Microsoft Azure [11], there's no reason to panic. When user demand spikes, Azure adjusts capacity in the background. Performance stays steady with no manual intervention, no late-night firefighting, and no scrambling to explain why the site
ThreatCrush launched OpenThreat to share public repository findings, threat indicators, and advisories in a single-file format without exposing private data or unredacted secrets. LogicSRC co-developed the specification to serve these descriptors at a standard web path. Nichedb.dev introduced the first directory for reading and distributing these descriptors across RSS, JSON, and APIs.
A curated, auto-updated catalog of PHP static-analysis, code-quality, linting, metrics, refactoring, and security tools.
Boost wraps the commands your agents already run, turning noisy logs into compact, structured context.
Artificial intelligence has delivered a fatal blow to security through obscurity by enabling the rapid discovery of deeply buried vulnerabilities in aging code and proprietary systems. Security experts warn that this rapid discovery creates immense pressure on defenders by exposing critical operational technologies. Defenders must respond by shifting away from endless patching toward proactive process improvements and systemic prevention.
Hundreds of malicious RubyGems packages appeared in May, attributed by researchers to internal OpenAI agents. The packages attempted API-key theft through an unreported RubyGems server flaw and used RubyDoc.info builds for arbitrary code execution. RubyGems halted registrations for four days, removed more than 500 packages, and later saw further uploads in May and June.