When a file is edited in a fork through the web editor, Gitea can first copy a branch from the base repository into the fork. That copy did not check whether the user may read the base repository's code. A user who had forked a repository and later lost access to it, for example after it was made private or they were removed as a collaborator, could keep copying its current branches into their fork and read them. The copy now requires read access to the code of the base repository.
Frappe india-payroll before 16.0.4 contains a missing authorization vulnerability in the get_employee_details, compute_tax_comparison and set_tax_regime methods of tax_regime_selector.py that allows authenticated low-privileged users to access other employees' payroll data. Attackers can supply another employee identifier to read base pay and earning and deduction components, or change the tax regime on draft Salary Structure Assignments to affect tax withholding.
The NLTK library's SteppingRecursiveDescentParser.parse() method performs an unbounded top-down search on ambiguous or self-embedding grammars without enforcing any time or step limits. This allows an attacker who can influence the grammar and input tokens to cause the parser to consume CPU indefinitely, resulting in a denial of service. The base RecursiveDescentParser enforces a max_time limit, but the stepping subclass overrides this behavior and lacks any timeout mechanism. This vulnerability affects NLTK versions prior to 3.10.4.
Scapy before 2.8.0 contains a denial of service vulnerability in BER_num_dec() and BERcodec_OID.do_dec() that allows remote attackers to exhaust CPU by supplying oversized OID arcs. Attackers can send ASN.1/BER data with many base-128 continuation bytes to cause quadratic CPU consumption or an uncaught ValueError crash on Python 3.11+.
A vulnerability in urllib3 used by Red Hat AI Base Images 3.5.8 (Spyre) allows a remote network attacker to intercept and modify HTTPS proxy traffic by exploiting improper TLS certificate validation. This occurs when the target server's TLS settings override the proxy's TLS configuration, disabling certificate verification on reused connections. Exploitation requires specialized application architectures using HTTPS proxies with customized TLS parameters. Default Red Hat deployments using direct endpoint communication or standard HTTP proxies are not affected. No fix is currently available for this issue in the affected images.
A double-free vulnerability exists in FFmpeg's NVIDIA NVDEC hardware decoder, affecting Red Hat AI Base Images 3.4.9 (Spyre). This flaw can be exploited by processing a specially crafted video file, potentially leading to memory corruption, arbitrary code execution, or denial of service in applications using FFmpeg with NVDEC acceleration. No fix is currently available from Red Hat, and mitigation options do not meet their criteria for deployment.
A security advisory from Red Hat addresses multiple vulnerabilities in Red Hat AI Base Images 3.5.7 (Spyre), including CVE-2026-11856, which involves libcurl incorrectly reusing Digest authentication headers across different HTTP origins. This flaw can lead to unintended information disclosure by sending authentication headers meant for one origin to another. The advisory notes no fixes are currently available for these issues in the affected AI base images. The libcurl vulnerability is fixed in version 8.21.0, affecting versions from 7.10.6 to 8.20.0. Mitigation involves correct usage patterns such as creating fresh handles for different origins or clearing authentication state before reuse.
CVE-2025-54503 is a vulnerability in AMD EPYC 9005 Series processors related to improper handling of overlapping protected memory ranges in the Secure Encrypted Virtualization (SEV) firmware. This flaw could allow a privileged attacker to misconfigure the Reverse Map Table (RMP), potentially leading to loss of memory integrity. The vulnerability has a medium severity with a CVSS 4.0 base score of 5.9. No specific affected versions or patches are provided in the available data.
CVE-2026-124999 is a path traversal vulnerability in braindecode versions 1.8.0, 1.8.1, and 1.8.2. It affects the EEGModuleMixin._from_pretrained function in the braindecode/models/base.py file, allowing remote attackers to manipulate the filename argument to traverse directories. The vulnerability has a medium severity with a CVSS 4.0 score of 5.3. Upgrading to braindecode version 2.0.0 mitigates this issue.
A security vulnerability has been identified in the apache-superset package related to the filelock component, which is a platform-independent file lock for Python. Versions prior to 6.1.0-r8 are affected. The vulnerability has a CVSS 3.1 base score of 6.1, indicating a medium severity issue. It involves a local attack vector with low complexity and requires low privileges but no user interaction. The impact includes low confidentiality, high integrity impact, and no availability impact. A patch is available in version 6.1.0-r8.
CVE-2026-96889 affects Red Hat AI Base Images 3.5.6 (Spyre). It is associated with vulnerabilities in the Google Brotli decoder, specifically an out-of-bounds write triggered by crafted Brotli-compressed streams. This flaw can lead to memory disclosure, denial of service, or potentially arbitrary code execution when decompressing untrusted Brotli data. Red Hat rates this issue as Important with high attack complexity due to required conditions for exploitation and platform hardenings that mitigate reliable code execution. No fix is currently available for the Red Hat AI Base Images 3.5.6 (Spyre).
Red Hat AI Base Images 3.4.7 (Spyre) are referenced in a security advisory that includes multiple CVEs, notably CVE-2026-111497, which involves an out-of-bounds write vulnerability in the Google Brotli decoder. This flaw can be triggered by crafted Brotli-compressed streams during decompression, potentially leading to memory disclosure, denial of service, or arbitrary code execution. Red Hat rates this issue as Important with high severity but not critical due to attack complexity and platform mitigations. No fixes are currently available for the Red Hat AI Base Images 3.4.7 (Spyre) as per the advisory. The advisory covers related Red Hat products including OpenShift AI. The vulnerability requires decompression of attacker-controlled Brotli data to be exploited.
A denial of service vulnerability exists in pyasn1 versions 0.6.4 and earlier used by smbclient-ng (>=3.1.0). The flaw causes quadratic-time CPU consumption when decoding constructed OCTET STRINGs or indefinite-length ANY types composed of many fragments, leading to CPU exhaustion. This affects BER and CER decoding paths and certain non-conforming DER inputs. No confidentiality or integrity impact is reported. The CVSS v3.1 base score is 7.5 (high severity).