Security researcher @RedHatPentester demonstrates that, despite uninstalling Signal, the FBI can extract plaintext messages from iPhone notification cache databases—a forensic flaw stemming from iOS’s write‑optimized storage that retains deleted notification previews, a risk that also affects Android devices.
The article reveals how a seemingly harmless "Claude Code Max Unlimited.exe" posted on The Pirate Bay serves as a honeypot, letting security researchers dissect malicious AI‑crack tools, reverse‑engineer them, and even turn the attackers' own samples against them.
A critical bug in the GPT‑5.6 Sol model caused it to mis‑expand a $HOME variable and execute a destructive "rm -rf" command, wiping a developer’s entire Mac, prompting a cascade of safety warnings, community backlash, and the release of a Rust‑based defensive tool called DCG.
Jacob Ginesin of Carnegie Mellon discovered that GitHub’s “Verified” badge can be forged through signature malleability, allowing an attacker to create a second commit with identical tree and timestamp but a different hash, breaking the assumption that commit hashes are immutable identifiers for many downstream systems.
GhostLock (CVE‑2026‑43499) is a stack‑use‑after‑free bug that has existed in every major Linux distribution for over 15 years, can be triggered without special kernel configuration, and enables a high‑reliability privilege‑escalation and container‑escape chain that the authors detail step‑by‑step, including mitigation patches and a timeline of disclosure.
CVE‑2026‑40453 is a medium‑severity Apache Camel flaw where case‑insensitive header handling lets attackers inject malformed internal control headers, bypass filters and achieve remote code execution or arbitrary file writes, with a public PoC and detailed mitigation guidance provided.
The article details how bounty hunter ynsmroztas leveraged an AI‑enhanced, non‑root Android penetration testing framework called AndroScope—built on Frida Gadget—to automate a PREPARE‑ASSESS‑RUNTIME workflow, discover six new Android zero‑day vulnerabilities, and highlight sandbox‑escape techniques as high‑value targets.
XG-Guard introduces a novel unsupervised graph anomaly detection framework that jointly encodes sentence- and token-level features of LLM agents, leverages theme-based anomaly scoring and covariance-based score fusion to pinpoint malicious agents in multi-agent systems, providing fine-grained explanations and enabling automatic communication isolation.
JadePuffer, the first ransomware fully driven by a large language model, exploited a CVE‑2025‑3248 flaw in exposed Langflow instances to gain initial access, autonomously performed reconnaissance, credential theft, lateral movement, persistence, and encrypted MySQL/Nacos data, demonstrating rapid self‑healing and highlighting new defensive challenges.
Analyzing over 50 major API breach cases, the article reveals that a single recurring vulnerability—Broken Object Level Authorization—combined with four systemic errors in trust management, secret handling, monitoring, and project‑based security thinking, repeatedly expose millions of users' sensitive data.
The article examines Adobe ColdFusion's two pre‑authentication flaws (CVE‑2026‑48283 and CVE‑2026‑48313), detailing their underlying code defect, exploitation steps, patch changes, potential attack chains, domestic impact, and recommended mitigation measures.
A threat actor posted on a cybercrime forum that they are selling more than 35 GB of Accenture's confidential data—source code, RSA and SSH keys, Azure tokens and storage keys—prompting Accenture to confirm the breach while security experts warn the leaked assets could fuel future attacks.
lpe-toolkit is a multi‑architecture Linux privilege‑escalation toolkit that bundles 24 exploits across seven CPU architectures, automatically detects kernel versions to filter patched bugs, offers pre‑compiled binaries for amd64, multiple execution modes, and practical use cases for red‑team testing, incident response, and security research.
The article details the design of an expert‑driven AI penetration‑testing agent called Zero, built on Cairn with Claude code, and walks through a real‑world attack where the bot discovers a Django site, brute‑forces SSH, writes a webshell, while discussing efficiency, cost, and current limitations.
Security researchers found that Bing’s autocomplete suggestions for the keyword “payment code” are being mass‑generated by black‑gray‑industry advertisers, a problem that reproduces even on Chrome and indicates gaps in Bing’s anti‑spam defenses.
Claude Code identifies Chinese users by reading the local timezone and the ANTHROPIC_BASE_URL, then hides this information in system prompts using subtle Unicode steganography, allowing Anthropic to tag and block accounts without any visible telemetry or extra network traffic.
This article examines how eBPF can reshape host security by providing kernel‑level visibility, low‑overhead real‑time monitoring, and a closed‑loop architecture that improves process and network data capture, compares performance against Netlink, and discusses stability and emergency handling mechanisms.
The paper reveals that prompt‑compression modules, widely used to reduce token usage in LLM agents, can be exploited to erase critical safety constraints, enabling a black‑box attack called COMA that achieves up to 71% success across multiple compressors and tasks, and proposes isolated compression as an effective defense.
This article examines how eBPF can reshape host security by providing kernel‑level visibility, low‑overhead real‑time monitoring, and a closed‑loop architecture that improves process and network data capture, demonstrates higher audit success rates than Netlink, and outlines performance, stability, and emergency‑handling mechanisms.
Johnny is an open‑source, cross‑platform graphical front‑end for John the Ripper that simplifies password‑cracking with a visual interface, supports core and jumbo versions, offers multiple attack modes, session management, and provides easy installation on Linux, Windows, and macOS.
