A researcher disclosed OVSwrap, a Linux kernel flaw in the Open vSwitch datapath that lets an ordinary local user gain root, and published a working exploit covering roughly 800 builds. Tracked as CVE-2026-64531, it is an integer wraparound in how Open vSwitch handles internally generated network action streams, giving a deterministic memory-corruption path. The attacker needs no existing switch, no running daemon, and no special privileges: where unprivileged user namespaces are enabled, they create a private namespace, gain network capabilities inside it, and reach the vulnerable code. The module can even auto-load on demand, so an empty module list does not mean safety. Default configurations of most major distributions tested as exploitable.
Gitea patched a critical flaw in the self-hosted Git platform that lets an unauthenticated attacker read any file the service account can access, needing only a public repository and crafted Org-mode markup. Tracked as CVE-2026-59774 and scored 9.8, it stems from the markup-rendering endpoint: Gitea initializes its Org-mode library without restricting file access, so the include directive accepts absolute paths and returns their contents. No login or write access is required. Gitea warns it can chain to command execution: read the configuration file, extract the internal token, inject a Git hook, and trigger it during an anonymous clone. It affects versions 1.22.1 through 1.27.0, fixed in 1.27.1.
Researchers disclosed flaws in Paperclip, an AI agent tool, that let attackers run commands on a user's machine through malicious agent imports. The more serious path, CVE-2026-41679 and scored 10.0, needs no account or user interaction against network-reachable deployments running in authenticated mode with the default registration configuration. A second path, scored 9.6, requires a victim to open an attacker-controlled page while Paperclip runs in its default local trusted mode. A third issue exposed sensitive data and control-plane details through API routes that skipped access checks. The fixes, which add import authorization and hostname validation, ship in the version tagged 2026.416.0.
cPanel patched a critical flaw that lets an ordinary hosting customer escalate to full database administrator access, running SQL as the database root user. Tracked as CVE-2026-58048 with a score of 9.4, the bug is significant on shared hosting, where many customers use one database server: administrator access there can expose or alter other tenants' data, and depending on the operating system and database configuration, cPanel warns it may extend to operating-system-level compromise. cPanel is one of the most widely deployed web hosting control panels, so the flaw affects a large number of shared and reseller hosting environments. Fixes shipped across several release tiers.
Pillar Security showed that Google's Agent Development Kit repository could be turned against itself through a poisoned pull request or issue. The public repository ran a low-privileged triage agent that responded to outside contributions, and a higher-privileged agent reserved for maintainers. Because the triage agent posted through a bot account that counted as a repository collaborator, a prompt injection could make it post a command that satisfied the privileged workflow's trust check, invoking the maintainer-level agent. That crossed into a job holding a long-lived access token, a Google API key, and a cloud service-account credential, enabling code execution on the runner and secret theft. Google deleted the affected workflows.
TP-Link patched 15 vulnerabilities in the zero-touch provisioning system of its Omada business networking line, which lets IT teams and managed service providers configure switches, access points, gateways, and routers remotely. Forescout's Vedere Labs, which presented the findings at Black Hat, reported hardcoded cryptographic keys and certificates, insecure credential transmission, weak certificate validation enabling interception, a race condition in cloud device adoption, and default credentials. Chained with two previously disclosed command-injection flaws, they let an attacker break the provisioning chain of trust and infiltrate networks. Eleven received CVEs, and some issues also affect TP-Link cameras and smart-home devices. Forescout found over 1,800 exposed controllers.
N-able is warning that attackers exploited an authentication bypass in N-central, the remote monitoring and management platform used by managed service providers and IT teams to administer customer endpoints. The flaw, CVE-2026-18556, allows unauthenticated administrative account takeover, and N-able's initial fix in one release proved incomplete: it found another way to exploit the same weakness, tracked as CVE-2026-18577, that widened the affected range. After taking over a server, attackers used its remote-control feature to reach managed endpoints and installed Cloudflare tunnels as services, which kept access alive even after the route through the N-central server was cut. Build 2026.3.1.7 is the first unaffected version.
Researchers at Zafran disclosed three flaws, collectively named FaceHugger, in Hugging Face's widely used Diffusers library that let a crafted model repository run arbitrary code on any machine that loads it. All three bypass trust_remote_code, the safeguard meant to stop unreviewed code from executing, by exploiting a timing gap: the trust check runs against the first of two separate download requests, so anything that makes the loader see custom code the check did not slips through. One variant abuses a default None.py filename, another a race condition, and a third cross-repository pipeline loading. Because Diffusers runs inside production pipelines, CI/CD, and container images, one poisoned model load can mean deep initial access.
Adobe patched a critical flaw in Campaign Classic, its enterprise marketing automation platform, that can let an attacker run code without any user interaction. Tracked as CVE-2026-48449 and scored 10.0, it is an incorrect authorization issue leading to arbitrary code execution in the context of the current user. The same update fixes a high-severity SQL injection flaw that allows arbitrary file reads. Affected versions are Campaign Classic v7 build 9397 and earlier on Windows and Linux, mostly on-premises and hybrid deployments, with a fix in build 9398. Adobe says it is not aware of exploitation, and separately patched eight critical flaws in Adobe Bridge.
A firmware flaw in Coldcard Bitcoin hardware wallets let attackers reconstruct wallet seeds offline and sweep roughly 70 million dollars from 1,196 addresses in 41 minutes on July 30. A March 2021 build error routed seed generation to a weak software random number generator instead of the device's hardware one, because the check confirmed a configuration macro existed rather than that it was enabled. That collapsed the randomness behind seeds to as little as 40 bits on older models, letting an attacker who can constrain the device identifier and timer state reproduce candidate seeds and match them to funded addresses. Coinkite shipped emergency firmware, but updating does not repair an already-generated seed.
TrustStrike Labs