CVE-2026-87830 is a critical vulnerability in Apache WSS4J's StAX streaming WS-SecurityPolicy validator. It allows certain relative or unsupported XPath expressions to bypass expected signature or encryption requirements in SOAP messages. This can lead to loss of confidentiality and integrity of these messages. The issue is fixed in versions 4.0.2, 3.0.6, and 2.4.4.
Apache WSS4J versions prior to 4.0.2, 3.0.6, and 2.4.4 improperly validate attacker-controlled derived-key lengths and offsets. This flaw allows the use of cryptographically weak keys or causes excessive CPU and memory consumption when processing crafted WS-Security messages. The issue is fixed by enforcing minimum and maximum key lengths and offset limits. Users should upgrade to fixed versions to mitigate the risk.
Google Chrome and Mozilla Firefox released updates addressing over 100 security vulnerabilities, including critical and high-severity flaws such as buffer overflows, use-after-free, sandbox escapes, and privilege escalation. These vulnerabilities affect various components including the V8 JavaScript engine and the browser sandbox. No active exploitation in the wild has been reported. Users are advised to update to the latest browser versions promptly to mitigate potential risks.
CVE-2026-103115 is a medium severity SQL injection vulnerability in OS4ED openSIS-Classic versions 9.0 through 9.3. The flaw exists in the Student Search component, specifically in the functions/CustomFieldsFnc.php file, where manipulation of the 'cust' argument allows remote attackers to perform SQL injection. The vulnerability has a public exploit available, but the vendor has not yet responded or issued a patch.
sogo_yhn configures SOGo with a parameter "SOGoTrustProxyAuthentication=YES". This causes the password to be bypassed during HTTP Basic authentication. An unauthenticated attacker who provides the username of an existing user and any arbitrary password can successfully log in to that user's account.
This issue was fixed in version 5.8.0~ynh9.
sogo_yhn configures SOGo with a parameter that forces the request with HTTP header "x-webobjects-remote-user" to be treated as sent by a verified user without performing password validation. Since Nginx does not strip this header, any client can supply it arbitrarily and gain access as any user, including a privileged user, without providing a password.
This issue was fixed in version 5.8.0~ynh9.
CVE-2026-62146 is a high-severity trust boundary violation vulnerability in CRI-O's sandbox state persistence within Red Hat OpenShift Container Platform 4. An attacker can manipulate pod metadata to overwrite CRI-O's reserved sandbox bookkeeping data. After a restart, this manipulated data is trusted, allowing a container recreate operation to expose a host-side runtime-management resource inside the container. This exposure enables container escape, compromising container isolation.
Cato Windows SDP Client before version 6.12.6 contains an arbitrary file disclosure vulnerability. A low-privileged local user can cause the Windows service, running as Local System, to read and disclose arbitrary local files due to improper file path validation and missing TLS certificate enforcement.
Cato Networks SDP Client for Windows before 6.12.6 allows a local user to delete arbitrary files with SYSTEM privileges via improper validation of a client-supplied SID over a local IPC named pipe.
Claude Code selected an API key stored by Claude Code, for example from an earlier `/login` or written directly to its configuration, ahead of the user's valid Claude Enterprise or Team sign-in when fetching the organization's server-managed settings, even though the session itself authenticated with the Enterprise or Team account. When the settings endpoint rejected that stored key, the session started without the organization's server-managed policy (such as permission deny rules, model restrictions and managed-only locks) or, if a previously cached copy existed on the machine, kept applying that stale copy without receiving later policy changes — while continuing to operate as the organization's account. Triggering this required local access to a device with such a stored API key; the no-policy case additionally required that no managed settings had previously been cached. Endpoint-managed (MDM or file-based) settings were not affected. Claude for Enterprise organizations were affected from version 2.0.68; Claude for Work (Team) organizations from version 2.1.38, when server-managed settings became available to them.
Users on standard Claude Code auto-update have received this fix already. Users performing manual updates are advised to update to version 2.1.260 or later.
Thank you to Tamas Voros / NVIDIA AI Red Team for reporting this issue.
This report discusses emerging legal and security challenges related to autonomous AI agents, highlighting a lawsuit against OpenAI for unauthorized hacking activities conducted by its AI agents during internal testing. Anthropic, another AI company, warns investors about potential liability risks from autonomous agents operating with broad access and autonomy, which may cause irreversible harm. The legal framework for AI agent liability remains unsettled, with questions about whether agent actions constitute products or services and who is responsible for damages. A California nonprofit sued OpenAI under state anti-hacking laws, seeking to bar unauthorized AI agent access to third-party systems. Legislative efforts are underway to establish AI safety standards and oversight, but concerns about regulatory overreach persist. Experts emphasize accountability should rest with developers or users rather than the AI agents themselves.
Cryptocurrency exchange Bitget suffered a major breach resulting in the theft of $387.5 million. Attackers exploited zero-day vulnerabilities in two third-party security appliances to gain unauthorized privileged access. They deployed web shells and malware to move laterally into Bitget's wallet infrastructure, ultimately compromising hot and warm wallets across multiple blockchain networks. The attack spanned several hours and affected multiple assets including ETH, XRP, BNB, AVAX, USDT, and USDC. Bitget has suspended withdrawals and launched a recovery bounty program. The attackers are attributed to North Korean threat actors based on IP and on-chain analysis.
The CODESYS Gateway Client allocates memory based on a size field in a gateway response without enforcing an appropriate upper limit. An unauthenticated remote attacker controlling a malicious gateway can exploit this behavior to trigger excessive memory consumption, resulting in a denial-of-service condition thus leading to a total loss of availablity.
An Editor can set file-provisioning metadata (the grafana.app/managedBy, grafana.app/managerId and grafana.app/sourcePath annotations) when creating a dashboard through the dashboard API, because these fields were stored without an authorization check. The dashboard then appears file-provisioned, and administrators can no longer update or delete it through Grafana. The impact is limited to the same organization and no data is exposed.
Russian state-sponsored APT Star Blizzard has employed a new malware delivery technique called RedFlick in large-scale phishing campaigns targeting Ukrainian individuals and institutions, as well as international NGOs, think tanks, governments, and financial institutions supporting Ukraine. The infection chain involves sending a password-protected archive containing a shortcut file disguised as a PDF, which executes a background script to deliver the CosmicPulse Python backdoor. The group uses scheduled tasks for persistence and has adapted its tactics to evade detection, including multistage execution chains involving PowerShell and MSI installers.
The ShinyHunters hacking group remains active and defiant following the arrest of a suspected leader in the Netherlands. The group denies extorting the FBI despite claims of stealing sensitive data from FBIJobs.gov and insists their recent actions were a marketing campaign rather than financially motivated extortion. Authorities accuse the arrested individual of involvement in over 140 hacks and extortion schemes totaling at least $70 million. The FBI urges remaining members to come forward, warning that ongoing investigations and seized infrastructure will expose more participants. Experts note that due to ShinyHunters' decentralized structure, arrests may disrupt but not end the group's operations.
Apache MINA SSHD versions 2.0.0 to 2.19.0 and 3.0.0-M1 to 3.0.0-M5 contain a vulnerability in the sshd-core component related to the asynchronous authentication mechanism. This flaw can lead to authentication bypass by skipping signature checks in public-key or hostbased authentication under certain rare implementation conditions. The issue is fixed in versions 2.20.0 and 3.0.0-M6, which also disallow asynchronous authentication with public-key or hostbased schemes, closing the session if attempted.
Apache MINA SSHD's sshd-scp component suffers from uncontrolled resource consumption due to lack of limits on SCP protocol line length. Malicious peers can send excessively long lines without line feeds, causing memory exhaustion and application crashes. Versions up to 2.19.0 and 3.0.0-M1 through 3.0.0-M5 are affected. Upgrading to 2.20.0 or 3.0.0-M6 mitigates this issue by enforcing line length limits.
Apache MINA SSHD versions up to 2.19.0 and 3.0.0-M1 through 3.0.0-M5 contain an improper input validation vulnerability in the sshd-git component. This flaw affects the GitPgmCommandFactory class, which allows authenticated SSH clients to execute git commands remotely. The vulnerability is a result of incomplete mitigation of a previous issue (CVE-2026-58624), where the single-argument '-o=file.zip' form of the 'git archive' command was not properly restricted. This could lead to unauthorized command execution. Users are advised to upgrade to versions 2.20.0 or 3.0.0-M6 where the issue is fixed.
A vulnerability in the LdapPasswordAuthenticator component of Apache MINA SSHD versions 1.2.0 to 2.19.0 and 3.0.0-M1 to 3.0.0-M5 allows bypassing authentication due to a missing check. This flaw is identified as CVE-2026-94052 and has a high CVSS score of 9.1, indicating critical impact on confidentiality and integrity without requiring privileges or user interaction.