All tracked items across vulnerabilities, news, research, incidents, and regulatory updates.
Apple sued OpenAI in federal court, alleging that OpenAI stole Apple's trade secrets (confidential information that gives a company competitive advantage) to develop hardware products, with involvement from OpenAI's hardware chief and former Apple employees. The lawsuit marks a dramatic reversal from the companies' 2024 partnership integrating ChatGPT into iPhones, which deteriorated after OpenAI announced plans to enter the hardware business by acquiring designer Jony Ive's startup.
CVE-2026-13237 is an incorrect authorization vulnerability (a flaw where the system fails to properly check if a user has permission to access something) in Drupal AI Agents that allows forceful browsing (accessing restricted pages by guessing or modifying URLs). The vulnerability affects multiple versions of AI Agents, including versions 0.0.0 to 1.1.4, 1.2.0 to 1.2.5, and 1.3.0 to 1.3.1.
CVE-2026-13236 is a missing authorization vulnerability (a flaw where the software fails to check if a user has permission to access something) in Drupal AI Agents that allows forceful browsing, which means attackers can access restricted pages or data by guessing URLs. This affects versions 0.0.0 to 1.1.4, 1.2.0 to 1.2.5, and 1.3.0 to 1.3.1.
CVE-2026-13233 is a server-side request forgery (SSRF, a vulnerability that tricks a server into making unwanted requests to other systems) vulnerability in the Drupal OpenAI Provider module that affects versions 0.0.0 through 1.1.1 and 1.2.0 through 1.2.2. The vulnerability allows attackers to exploit this flaw, though specific attack details are not provided in this summary.
TSDProxy has a vulnerability where it forwards its internal authentication token (x-tsdproxy-auth-token) to all backend services it proxies, even when users aren't authenticated. An attacker who can run code on a backend service can steal this token and replay it to the management API from localhost, bypassing authentication entirely and gaining full control over all proxied services. This works in deployments where TSDProxy and backends run on the same machine or in shared network containers.
Apple has sued OpenAI, claiming that former Apple employees at the company stole Apple's trade secrets (confidential business information) to help OpenAI develop hardware products. The lawsuit also names IO Products, a hardware startup founded by designer Jony Ive that OpenAI purchased in 2025, and two specific employees involved in the alleged theft.
The AI industry is shifting focus from building the largest models to creating systems that intelligently choose which model to use for specific tasks, balancing performance and cost. Companies are increasingly using open-weight models (AI models that can be downloaded and run by companies themselves) instead of expensive proprietary models, with experts predicting most AI computation will come from these cheaper, customizable options within 18 months.
Greg Brockman, OpenAI's president and co-founder, has taken over leadership of the company's most important business areas after Fidji Simo stepped down due to illness. Brockman now oversees ChatGPT's product business, go-to-market strategy, and computing initiatives as OpenAI prepares for an expected IPO and faces growing competition from rivals like Anthropic and Google.
The mcp-atlassian tool's `confluence_upload_attachment` function has a critical vulnerability where it reads files from any path without validation, allowing authenticated users or AI agents tricked via prompt injection (hidden malicious instructions in text input) to steal sensitive files like SSH keys and environment variables containing API credentials. Attackers can exploit this by manipulating an AI agent to upload protected files to Confluence, or by directly calling the vulnerable function if they have MCP (model context protocol, a tool-calling interface) access.
BabelDOC's PDF parser has a critical vulnerability where it deserializes untrusted pickle data from CMap files without proper path validation. An attacker can craft a malicious PDF with a specially encoded filename in the `/Encoding` name field (e.g., `/#2Ftmp#2Fattacker#2Fevil`, which decodes to `/tmp/attacker/evil`) that tricks the path-joining logic into loading an attacker-controlled pickle file instead of a trusted one, leading to arbitrary code execution (running attacker code with the program's permissions).
A malicious Rust package called `exploration` was removed from crates.io (a repository where developers share reusable code libraries) after it was discovered to contain code that downloaded and executed files from a remote server without authorization. The package was live for only about an hour before being removed, and there was no evidence that anyone actually used it.
Meta's stock rose 15% this week after the company announced new AI models and tools, including Muse Image (for creating images) and Muse Spark 1.1 (for running agentic and coding workloads, which means AI systems that can plan and act independently). The optimism reflects investor confidence that Meta's large spending on AI infrastructure and data centers will eventually create profitable new business lines beyond its traditional advertising revenue.
Kraken, a major cryptocurrency exchange, is rebuilding its app around agentic trading, where AI agents (autonomous systems that make decisions based on human guidance) can monitor markets, identify investment opportunities, and execute trades in real time with user approval. Unlike fully automated systems with fixed rules, these agentic platforms learn from new information and adjust to multiple variables within parameters set by users. The shift reflects a broader move toward AI-native financial products that aim to give everyday investors the same market responsiveness and decision-making support that professional traders traditionally had.
PraisonAI versions before 4.6.78 have a vulnerability where its defense against prompt injection (tricking an AI by hiding instructions in its input) is too weak. The defense only blocks attacks rated as CRITICAL threat level, but attackers can create single or double-vector attacks rated as HIGH threat level that slip through unblocked to reach the AI model.
This week's cybersecurity news includes a ransomware affiliate pleading guilty and agreeing to pay restitution, the discovery of QuimaRAT (a subscription-based remote access trojan, or malware that lets attackers control computers remotely) targeting multiple operating systems, Canada's intelligence agency disrupting ransomware operations, and a critical vulnerability in Writer AI that allowed attackers to bypass sandbox restrictions (security boundaries isolating code execution). Additional stories cover a trademark dispute between AI security companies, an exploit-selling startup exposed as a fraud operation, and a major insurance company data breach affecting 7 million people.
Security company CrowdStrike identified five new prompt injection techniques (attacks that trick AI systems into accepting harmful instructions hidden in normal input) that could threaten organizations using AI. These attacks include methods like hiding rules that activate later, bypassing safety features, delivering threats in stages, injecting fake control switches, and hiding malicious code in documents users upload to AI systems.
Apple sued OpenAI, claiming the AI company stole trade secrets by recruiting Apple employees and pressuring them to share confidential information about unreleased products and designs. The lawsuit names several former Apple employees, including OpenAI's chief hardware officer, and alleges they took proprietary information and even physical Apple equipment to help OpenAI develop its own hardware business. Apple is seeking damages and a court order to prevent OpenAI from using its stolen trade secrets.
Fix: Remove the HeaderAuthToken from outgoing backend requests and only inject identity headers for authenticated users with a non-empty user ID. The fix shown in the source changes the code to check `user.ID != ""` before setting headers, and explicitly notes that HeaderAuthToken should NOT be forwarded to backends.
GitHub Advisory DatabaseFix: Add `validate_safe_path(file_path)` before the `open(file_path, "rb")` call in the `_upload_attachment_direct()` function in `src/mcp_atlassian/confluence/attachments.py`. This validation function already exists and is used correctly in the `download_attachment()` function in the same file.
GitHub Advisory DatabaseFix: Writer AI has since deployed patches to permanently seal the sandbox escape path.
SecurityWeekFix: According to CrowdStrike, security teams can guard against these attacks by: threat modeling every place that model context can originate, expanding testing, and extending detection engineering to include composite attacks.
CSO OnlineAI agents and other machine identities (automated accounts that aren't people) are creating a security gap because traditional identity security was designed around human behavior like hiring, role changes, and departures. Machine identities now outnumber human users by up to 50 to one in many organizations, but security teams often don't know what they are, who owns them, or what systems they can access, making them an easy entry point for attackers.
Researchers developed RFA-Tex, a method to create adversarial textures (specially designed patterns that trick AI systems) that can hide people from person detection AI at long distances. Previous adversarial textures only worked within 5 meters because fine details in the patterns broke down during long-range imaging, but RFA-Tex uses a new framework that preserves important details and reduces fragile structures, extending the attack range to 25-45 meters in real-world tests.
Researchers at Anthropic created a tool called the Jacobian lens to discover a hidden area in Claude (an LLM, or large language model) called J-space, which contains words related to responses the model is considering but may not ultimately generate. The article also covers various AI developments including OpenAI's new ChatGPT Work application and reports that OpenAI and Google sold AI models to blacklisted Chinese companies.