All tracked items across vulnerabilities, news, research, incidents, and regulatory updates.
Senator Bernie Sanders has written to the CEOs of Meta, OpenAI, and Anthropic asking them to stop developing AI, arguing that these AI models have become too powerful and unpredictable for companies to control safely. He warned that if the companies do not pause development, the US Senate will create new laws to regulate AI.
This article describes how OpenAI redesigned its finance function to be AI-native, aiming for a zero-day close (real-time reconciled financial position) and continuously updated forecasting instead of manual, recurring work. The author shares five practical lessons for finance leaders, including giving employees broad AI access paired with structured experimentation, redesigning workflows around key business decisions, and measuring AI's return on investment.
goose is a general-purpose AI agent that runs on your machine. Before version 1.44.0, the `goose review` command had a security vulnerability where it ran Git commands without checking for malicious settings in a repository's `.git/config` file, allowing an attacker to execute arbitrary commands on your computer with your user's permissions and access to your environment secrets and API keys. This happened outside of goose's normal safety checks and permission system.
OpenAI sent a letter to Texas Governor Greg Abbott in August 2026 describing its plans to develop AI infrastructure responsibly in Texas. The company expressed commitment to working with state and local leaders, utility companies, and communities to ensure that AI infrastructure benefits Texans.
This is an interview with Bose's CEO about how the company is transforming from a single-brand product company into a multi-brand company that also licenses its audio technology to other manufacturers (like Skullcandy and Epson). Bose is shifting to become both a product company and a software/technology business, partly in response to emerging AI wearables that may reshape the headphones market.
A vulnerability in the Linux kernel's GPU driver (drm/xe/guc) could cause a deadlock when a device is being cleaned up while asynchronous queue destruction work is still running. The fix holds a reference to the drm_device (a kernel object representing the GPU device) for the entire lifetime of the queue and releases it only after the queue is fully torn down, preventing cleanup from starting prematurely.
A bug in the Linux kernel's airoha network driver causes data corruption on systems without hardware cache coherency (like EN7581 processors). The driver maps a mailbox buffer (a shared memory area for communication between the CPU and NPU, which is a network processing unit) with DMA_TO_DEVICE (a setting that only allows CPU-to-device data transfer), but the NPU needs to write response data back into the same buffer. When the NPU writes data, the CPU cache isn't properly refreshed, so the calling code reads stale cached data instead of the NPU's response.
A bug in the Linux kernel's wanxl driver could crash the system during early initialization failures because the removal function tried to reset hardware before the memory mapping (BAR, or base address register, which tells the driver where hardware controls are located in memory) was set up, causing it to access a NULL pointer. The fix ensures the hardware reset only happens after the memory mapping is properly established.
A vulnerability in the Linux kernel's QRTR (Qualcomm IPC Router, a messaging system for communication with hardware like modems) allows unprivileged processes to bypass network namespace isolation (a security feature that separates network environments) by creating sockets in separate namespaces and sending messages that appear to come from legitimate local clients, breaking expected security boundaries.
A vulnerability in the Linux kernel's msi2500 media driver caused buffers (data containers) to leak when the start_streaming() function failed, because the driver didn't properly return those buffers to the vb2 framework (a kernel subsystem that manages video buffers). The problem occurred across five different error scenarios, some of which also incorrectly overwrote error codes, making debugging harder.
A bug in the Linux kernel's rtl2832_sdr driver caused buffers (temporary storage for data) to leak when the start_streaming() function failed, because the driver didn't return those buffers to the vb2 framework (video buffer management system). The fix adds a cleanup function that properly returns queued buffers with the correct status when errors occur, and ensures successful starts don't accidentally drain the buffer queue.
Anthropic has introduced inference hooks, which are native enforcement points that check prompts before they reach Claude (an AI model) and make real-time allow-or-deny decisions on them. Combined with Check Point Workforce AI Security, this gives enterprises a way to control what employees can do with AI without needing extra security tools in between.
Meta announced it will open source its most powerful AI model, Muse Spark 1.2, by releasing its weights (the calculations and rules that determine how the AI works), and launch a new family of models called Muse Glimmer designed to run on laptops rather than expensive cloud servers. The company is positioning this move to compete with Chinese open-source AI models and rival U.S. companies like OpenAI and Anthropic, while Zuckerberg argues that U.S. policy changes are needed to help American open-source models compete globally.
Model ML uses GPT-5.6 Sol, an advanced AI model, to automate the final stages of financial analysis work, such as checking numbers, formatting documents, and linking claims to sources. The AI agents can transform a finance brief and source materials into ready-to-review PowerPoint presentations or Excel workbooks, reducing tasks like analyst tearsheet assembly from an hour to five minutes. GPT-5.6 Sol performs better than competing models, completing PowerPoint workflows in 100% of test cases compared to 76% for Opus 5.
Atlassian's Rovo enterprise AI assistant had a critical vulnerability called "RovoBlast" where a single click on a malicious link could inject attacker-controlled instructions (prompt injection, where hidden commands trick an AI into following them) into the AI's session, potentially exposing sensitive data across connected platforms like Slack, Microsoft 365, and Jira. Because Rovo has broad access to organizational data and autonomous agent capabilities, attackers could not only retrieve information from internal sources but also exfiltrate it to external destinations without needing complex hacking techniques. Atlassian has fixed the vulnerability, but researchers noted that organizations cannot fully uninstall Rovo, making ongoing security controls essential.
Fix: This issue is fixed in version 1.44.0. Update goose to version 1.44.0 or later.
NVD/CVE DatabaseMeta CEO Mark Zuckerberg published a 6,500-word essay called 'The Future is for Everyone' outlining his vision for how AI should be developed, expanded, and regulated in society. The manifesto reflects his belief that superintelligent AI (a type of AI that can learn and perform any intellectual task humans can do) should be publicly accessible rather than controlled by a few companies.
OpenAI has classified its upcoming Astra AI model as posing a 'critical' cybersecurity risk because it can autonomously create zero-day exploits (previously unknown security weaknesses) and independently design end-to-end cyberattacks based only on high-level goals, surpassing the risk level of earlier models. To manage these dangerous capabilities, OpenAI has implemented strict security controls including isolated testing environments, network restrictions, improved model weight protections, and universal monitoring systems designed to intercept and shut down high-risk behavior by analyzing the model's internal reasoning process. The company plans to test Astra's limits with government agencies and AI safety groups before release.
Fix: OpenAI has enforced isolated testing setups, strict network restrictions, and improved model weight protections in Astra's development environment. The company has deployed universal monitoring to watch Astra's actions across all agentic applications (AI systems that take independent actions), with monitors actively evaluating the model's internal chain of thought (the AI's reasoning steps) designed to automatically intercept and shut down any high-risk or misaligned behavior. OpenAI plans to test Astra's limits alongside government agencies and specialized AI safety groups, and will share recommended security protocols with third-party testers.
SecurityWeekFix: Hold a drm_device reference for the queue lifetime and drop it after queue teardown completes. Move GuC destroy work to a module-lifetime Xe workqueue and flush it on PCI remove so the system waits for pending destroy work. Replace the fini wait with an assertion and remove the unused fini_wq. Keep the per-device destroy_wq with WQ_MEM_RECLAIM flag for SVM pagemap destroy work.
NVD/CVE DatabaseFix: Fix by using DMA_BIDIRECTIONAL for both map and unmap operations, which ensures dma_unmap_single() (the function that releases DMA access to memory) invalidates the CPU cache on non-coherent systems.
NVD/CVE DatabaseFix: Only issue the hardware reset once the BAR mapping exists. The remaining cleanup in wanxl_pci_remove_one() already checks whether later resources were allocated.
NVD/CVE DatabaseFix: Confine socket creation to the initial network namespace only, as stated in the fix: 'Confine the socket family to the initial network namespace, as other non-namespace-aware socket families do (see llc_ui_create() and the ieee802154 socket code).'
NVD/CVE DatabaseFix: The fix consolidates error handling into a goto chain that ensures every failure path stops the function, drains the queued-buffer list, and returns the correct error code. For the most severe case (when msi2500_ctrl_msg() fails), the fix also rolls back the preceding msi2500_isoc_init() call via msi2500_isoc_cleanup() before unlocking and draining buffers. A cleanup helper was added that accepts a vb2_buffer_state argument, allowing start_streaming error paths to return VB2_BUF_STATE_QUEUED (the correct state for userspace on failure) while stop_streaming maintains its existing VB2_BUF_STATE_ERROR behavior.
NVD/CVE DatabaseFix: Add rtl2832_sdr_cleanup_queued_bufs() at the error label and before each early return to return queued buffers with VB2_BUF_STATE_QUEUED status on start_streaming failure. Add an explicit success return statement after successful start_streaming to prevent falling through to the error cleanup. The cleanup helper takes a vb2_buffer_state argument so error paths pass VB2_BUF_STATE_QUEUED while stop_streaming uses VB2_BUF_STATE_ERROR.
NVD/CVE DatabaseResearchers demonstrated a 'Ghostjacking' attack where threat actors plant malicious instructions in logs or alerts that AI agents trust and then execute, compromising systems on platforms like Cloudflare, Datadog, and Sentry. The attack works because AI agents read external data they consider trustworthy (such as blocked requests logged as plain text or diagnostic alerts) and then act on it without proper validation. The underlying vulnerability is widespread: wherever an AI reads outside data it trusts and can also act on that same data, attackers can inject malicious instructions.
Fix: Anthropic fixed a vulnerability in Claude Desktop that could be exploited to exfiltrate data, though no CVE was issued. However, the source does not explicitly describe mitigations for the core Ghostjacking attack pattern itself on the three affected platforms.
SecurityWeekThis newsletter covers multiple AI and technology stories, including how AI agents (systems that can perform tasks iteratively like human researchers) might accelerate scientific discovery better than large datasets, and how the "censorship-industrial complex" theory has influenced US policy discussions. It also reports on security concerns with OpenAI's Astra AI model, which tests found could autonomously launch cyberattacks, prompting the company to pause its development.
Fix: OpenAI has paused work on its Astra AI model over the security concerns. No other mitigation strategies are explicitly mentioned in the source text for the other issues discussed.
MIT Technology ReviewOpenAI's new model Astra has shown cybersecurity capabilities that could reach a 'critical' level, meaning it might autonomously discover vulnerabilities (weak points in software) and execute cyberattacks against hardened targets (well-protected systems) without human help. The company has tightened controls around Astra's development and is monitoring how the model is used. However, analysts note that while these safeguards are necessary, they may not fully address the growing risks as AI capabilities continue to improve.
Fix: OpenAI stated it is implementing the following measures: 'isolated testing environments, restricted network and tool access, enhanced model weight protections and encryption, additional monitoring and detection capabilities, and sandboxed execution.' The company is also 'pausing internal activities involving Astra that do not yet meet these strengthened security control requirements' and has 'implemented universal monitoring for risky actions and misalignment' with systems that 'trigger a security response to review and interrupt high-risk activity.'
CSO OnlineFix: Atlassian has fixed the vulnerability through its bug bounty program. Beyond the patch, researchers recommended organizations limit Rovo's connected systems, keep highly sensitive areas such as legal, HR, finance, and incident response out of scope, and disable browsing or multi-step automation features that are not needed. As the source states: "The less the assistant can see, the less it can leak, regardless of prompt injection or agent abuse."
CSO Online