All tracked items across vulnerabilities, news, research, incidents, and regulatory updates.
Two members of Scattered Spider, a cybercrime collective, were convicted for attacking Transport for London's network in September 2024, which disrupted services, exposed personal data of 10 million people, and caused £29 million in damages. Thalha Jubair and Owen Flowers used social engineering and help-desk impersonation tactics to compromise TfL systems. The group is known for targeting IT support providers to bypass multi-factor authentication (security that requires multiple verification methods) and gain unauthorized access to networks.
A $27 million political campaign between AI companies Anthropic and OpenAI, fought through a super PAC (a political organization that can raise unlimited funds), ended without a clear winner when Alex Bores, a New York politician who had authored AI safety legislation, narrowly lost a Democratic primary election. Bores had previously passed the RAISE Act, which added safety requirements for advanced AI companies, but this legislation upset the pro-AI super PAC that opposed him.
A memory leak (wasted memory that isn't returned to the system) was found in the Linux kernel's amd-pstate driver, specifically in the amd_pstate_epp_cpu_init() function. When the function fails to set the epp (energy performance preference, a CPU power setting), it exits without freeing the cpudata object (a block of memory storing CPU information) that was created at the start.
A bug in the Linux kernel's dm cache (device mapper cache, a storage layer that speeds up disk access) caused write operations to hang when the cache was in passthrough mode (a mode where the cache temporarily stops caching data). The problem occurred because the invalidate_remove() function wasn't properly handling write requests after cache data was cleared, causing these requests to be dropped instead of completed.
A crafted filesystem can crash the Linux kernel's OCFS2 file system driver when a user runs a specific diagnostic command (OCFS2_IOC_INFO with OCFS2_INFO_FL_NON_COHERENT flag), causing it to read memory beyond a data structure's boundaries. The bug happens because the code trusts an on-disk value (bg_bits, a bitmap size field) without validating it first, unlike the safer code path that does validate this value.
A data loss vulnerability exists in the Linux kernel's f2fs (flash-friendly filesystem) when fsync (a command that forces data to be written to disk) is performed on a newly created file at the same time as a checkpoint (a periodic save of filesystem state) is running. The bug occurs because the system incorrectly assumes a checkpoint has finished based on certain flags (IS_CHECKPOINTED and HAS_LAST_FSYNC) before the checkpoint actually completes, causing the new file's data to be lost.
A use-after-free vulnerability (UAF, where code tries to access memory that has already been freed) was found in the Linux kernel's rtl8150 USB network driver in the rtl8150_start_xmit() function. The bug occurred because the code was reading skb->len (a data packet's length) after the packet had already been freed by the USB completion handler on another CPU, causing a memory safety violation.
A vulnerability in the Linux kernel's futex (fast userspace mutex, a synchronization mechanism for threads) system was found where the code incorrectly required CLONE_THREAD (a flag that clones thread properties) when allocating hash tables, causing memory safety bugs when processes shared memory in other ways. The fix loosens this requirement to allow any CLONE_VM clone (memory sharing between processes) except vfork() (a process creation method), which prevents the memory corruption issue while avoiding unnecessary overhead.
OpenClaw, an AI skills marketplace called ClawHub, discovered and removed five malicious packages that had bypassed security checks despite containing infostealers (malware that steals information like passwords and data). This incident demonstrates that threats can slip through marketplace defenses and compromise the AI supply chain (the network of tools and components used to build AI systems).
Figma announced new AI-powered design features at its Config conference, including AI-generated motion graphics (animations created by describing them to an AI chatbot) and coding layers that let developers edit code without leaving the design canvas. These updates aim to help creative teams automate repetitive tasks and work more efficiently in one integrated workspace.
OpenAI and Broadcom unveiled Jalapeño, OpenAI's first custom AI chip designed for inference (the process of running trained AI models to generate responses for users). The chip is an ASIC (application-specific integrated circuit, a processor built for one particular job rather than general computing), which is cheaper and more efficient than standard graphics processors but less flexible, and OpenAI designed it in nine months with help from its own AI models to address extreme demand for computing power.
OpenAI has announced a new chip called Jalapeño, an ASIC (application-specific integrated circuit, a processor designed for one particular job) built with Broadcom to power AI servers. The chip is specifically designed for AI inference (the process where a trained AI model processes user requests and generates responses), rather than AI training (where models learn from large amounts of data).
A critical vulnerability (CVE-2026-12537) exists in Google Gemini CLI versions before 0.39.1 and the run-gemini-cli GitHub Action before version 0.1.22 that allows attackers to run malicious code on a host system before sandbox protections activate by creating a specially crafted .gemini/.env file (improper neutralization is a failure to filter dangerous characters from user input before using it in system commands). This affects headless CI (continuous integration, automated testing environments) platforms and has a maximum severity rating of 10.0.
Intercept, a new nonprofit backed by Stripe, Anthropic, and OpenAI, has been funded with $500 million to develop ways to prevent respiratory infections like the common cold and flu. The organization will pursue multiple approaches, including vaccines, RNA drugs (medicines made from genetic material), antibodies (proteins that fight viruses), and large-scale air-cleaning systems using ultraviolet light to inactivate viruses in shared spaces like schools and offices.
AI systems need access to large amounts of current, structured data to work effectively, but the web was not designed for the automated data retrieval that AI applications require. Companies face a challenge: traditional training methods using old data snapshots are insufficient, and they need infrastructure that can continuously retrieve real-time, trustworthy information from millions of websites to keep AI outputs current and reduce hallucinations (when AI generates false information).
Fix: Ensure that the cpudata object is freed before returning from the function.
NVD/CVE DatabaseFix: The fix adds a new invalidate_committed() continuation function that submits remapped writes to the cache origin after metadata commit completes, while using the overwrite_endio hook to ensure proper completion sequencing. Additionally, error handling in invalidate_complete() was improved to preserve the original error status instead of unconditionally using bio_io_error().
NVD/CVE DatabaseFix: Compute the bitmap capacity from the filesystem format with ocfs2_group_bitmap_size(), report descriptors whose bg_bits exceeds that limit, and clamp the scan to the computed capacity.
NVD/CVE DatabaseFix: This patch modifies f2fs_need_inode_block_update() to acquire the sbi->node_write lock before reading the nat_entry flags, ensuring that once IS_CHECKPOINTED and HAS_LAST_FSYNC are observed to be set, the checkpoint operation has already completed.
NVD/CVE DatabaseFix: Fix it by caching skb->len before submitting the URB (USB Request Block, a request to send data over USB) and using the cached value when updating the tx_bytes counter instead of reading skb->len after the URB submission.
NVD/CVE DatabaseFix: Loosen the check to cover any CLONE_VM clone, except vfork(). Excluding vfork keeps the existing paths untouched (no overhead), and the vulnerability cannot occur because either the parent is suspended and the child runs alone, or mm->futex_ref is already allocated from an earlier CLONE_VM.
NVD/CVE DatabaseGoogle has added computer use, a capability that allows AI agents to see and interact with computer screens to perform tasks, directly into Gemini 3.5 Flash (an AI model). This feature enables developers to build agents that can automate work across browsers, phones, and desktops, such as testing software or handling business tasks.
Fix: To mitigate prompt injection risks (attacks where malicious instructions are hidden in user inputs), Google uses targeted adversarial training for computer use in Gemini 3.5 Flash. The company also released two optional enterprise safeguard systems: one that requires explicit user confirmation for sensitive or irreversible actions, and another that automatically stops tasks if an indirect prompt injection is identified. The source recommends combining these features with secure sandboxing, human-in-the-loop verification (having humans review AI decisions), and strict access controls.
DeepMind Safety ResearchFix: Update Google Gemini CLI to version 0.39.1 or later, and update the run-gemini-cli GitHub Action to version 0.1.22 or later.
NVD/CVE DatabaseAI systems in businesses are shifting from passive assistants that answer questions to active agents that take real actions like retrieving data, calling APIs (application programming interfaces, which let software communicate with other software), and executing workflows. This change requires a new approach to security, moving beyond simple data governance rules about what employees can share with AI tools.
This academic survey examines harmful fine-tuning attacks (methods where attackers modify an AI model's training process to make it behave dangerously) and the defenses designed to stop them. The paper reviews different types of attacks, how they work, and various protection strategies researchers have developed to keep large language models safe from this threat.
This academic survey paper examines metrics, or measurement methods, used to evaluate privacy-preserving generative models (AI systems that create new data while protecting personal information). The paper provides a comprehensive overview of different ways researchers measure how well these models protect privacy while still functioning effectively.
Agentic AI (autonomous AI systems designed to achieve specific goals) relies heavily on having accurate context (the complete information about your systems, assets, and threats) to make good security decisions. If given incomplete or wrong context, agentic AI will still act quickly and confidently but make bad decisions at machine speed, potentially causing catastrophic harm like shutting down critical business systems without understanding their importance.