✨ The Containment Era is here. Secure AI workloads before they breach. →The Containment Era is here. →The Containment Era is here. →Explore ✨
Computer Software/Engineering
Breach intelligence, attack campaigns, and threat reports targeting the Computer Software/Engineering sector.
Explore Other Sectors
Computer Software/Engineering Threat Reports
DifyTap Vulnerabilities: A Wake-Up Call for AI Platform Security
In June 2026, security researchers identified four critical vulnerabilities, collectively termed 'DifyTap,' in the Dify AI platform. These flaws—CVE-2026-41947, CVE-2026-41948, CVE-2026-41949, and CVE-2026-41950—allowed unauthorized access to sensitive data, including private AI chat histories and documents across tenants. Exploitation could lead to significant data breaches and compromise of AI applications. The DifyTap vulnerabilities underscore the escalating risks associated with AI platforms, emphasizing the need for robust security measures and prompt patch management to protect sensitive information and maintain trust in AI-driven services.
1 month ago
Kill Chain
GitHub Actions Enhances Security with 'actions/checkout' v7 Update
In June 2026, GitHub enhanced the security of its software supply chain by updating the 'actions/checkout' action to block common 'pwn request' attack patterns. These attacks exploit the 'pull_request_target' workflow trigger to execute malicious code with full workflow privileges. Effective June 18, 2026, 'actions/checkout' version 7 and later versions refuse to fetch fork pull request code in 'pull_request_target' and 'workflow_run' workflows by default, unless explicitly configured otherwise. This change aims to prevent unauthorized code execution and protect repositories from potential compromises. This update is particularly relevant now due to the increasing prevalence of supply chain attacks targeting CI/CD pipelines. By enforcing stricter defaults, GitHub addresses a critical vulnerability that has been exploited in recent incidents, thereby strengthening the overall security posture of the developer community.
1 month ago
Kill Chain
Guarding AI Memory: Microsoft's Comprehensive Security Approach
In June 2026, Microsoft published an article titled 'Guarding AI Memory,' highlighting the critical importance of securing AI memory systems. The article discusses how AI memory transforms AI systems from stateless tools into learning collaborators, thereby increasing their attack surface. It emphasizes that without memory, attackers need to achieve their objectives in a single prompt, whereas with AI memory, they can shape behavior gradually over time or plant memories that influence agent reasoning after the original context is gone and user awareness is lower. Microsoft outlines a defense-in-depth approach to protect AI memory, spanning every layer of the stack: storage, retrieval, model interaction, and user control. This approach includes implementing sanitization checks on memory writes, governing stored memories with existing data policies, and providing observability through audit logs. The article also presents a guiding framework for building safe AI memory, emphasizing principles such as establishing intent and provenance before persistence, enforcing boundaries outside the model, treating retrieval as a risk decision, providing full lifecycle visibility for security teams, and keeping users in control. This publication underscores the evolving threat landscape in AI systems and the necessity for robust security measures to protect AI memory from potential attacks.
1 month ago
Kill Chain
OpenAI's 'Patch the Planet' Initiative: A New Era in Open-Source Security
In June 2026, OpenAI, in collaboration with Trail of Bits, launched 'Patch the Planet,' an initiative aimed at enhancing the security of critical open-source software. This program pairs OpenAI's advanced AI models, such as GPT-5.5-Cyber, with human security engineers to identify vulnerabilities, develop patches, and assist maintainers in integrating these fixes. Early participants include projects like cURL, Go, Python, Sigstore, and pyca/cryptography. The initiative has already led to the discovery of hundreds of security issues and the merging of numerous patches, significantly improving the security posture of these foundational software components. ([techcrunch.com](https://techcrunch.com/2026/06/22/openai-launches-new-initiative-to-help-find-and-patch-open-source-bugs/?utm_source=openai)) The relevance of this initiative is underscored by the increasing reliance on open-source software in critical infrastructure and the persistent challenges in maintaining its security. By combining AI-driven analysis with expert human intervention, 'Patch the Planet' addresses the pressing need for scalable and efficient vulnerability remediation in the open-source ecosystem.
1 month ago
Kill Chain
Anthropic's Fable 5 AI Model Compromised Shortly After Release
In June 2026, Anthropic released Fable 5, a publicly accessible AI model designed with safety classifiers to prevent misuse in areas like cybersecurity. Despite extensive pre-release testing, researchers identified methods to bypass these safeguards, enabling the model to generate potentially harmful content. This led to the U.S. government imposing export controls on Fable 5 and its more advanced counterpart, Mythos 5, citing national security concerns. The incident underscores the challenges in securing advanced AI systems against unintended applications. The rapid circumvention of Fable 5's safety measures highlights the evolving nature of AI vulnerabilities and the necessity for continuous monitoring and adaptive security protocols in AI development.
1 month ago
Kill Chain
Microsoft's 'AutoJack' Vulnerability: A Wake-Up Call for AI Agent Security
In June 2026, Microsoft disclosed a critical vulnerability chain, dubbed 'AutoJack,' in its AutoGen Studio—a tool for developing AI agents. This flaw allowed malicious web pages to exploit AI agents' web browsing capabilities, leading to remote code execution (RCE) on the host system. The attack combined three weaknesses: the AI agent's browser being treated as a trusted 'localhost' source, lack of authentication on the Model Context Protocol (MCP) WebSocket, and the ability to execute arbitrary commands via manipulated URL parameters. ([csoonline.com](https://www.csoonline.com/article/4187155/microsoft-says-web-enabled-ai-agents-can-trigger-host-level-rce.html?utm_source=openai)) The 'AutoJack' incident underscores the evolving security challenges in AI agent frameworks, highlighting the need for robust authentication and authorization mechanisms, especially when agents interact with untrusted web content. Organizations must reassess their security postures to address these emerging threats. ([microsoft.com](https://www.microsoft.com/en-us/security/blog/2026/05/07/prompts-become-shells-rce-vulnerabilities-ai-agent-frameworks/?utm_source=openai))
1 month ago
Kill Chain
Critical 'PixelSmash' Vulnerability in FFmpeg's MagicYUV Decoder (CVE-2026-8461)
In June 2026, a critical vulnerability known as 'PixelSmash' (CVE-2026-8461) was identified in FFmpeg's MagicYUV decoder, affecting versions prior to 8.1.2. This heap out-of-bounds write flaw allows attackers to execute arbitrary code or cause denial-of-service conditions by tricking users into opening malicious AVI, MKV, or MOV files. Applications utilizing FFmpeg's libavcodec, such as Jellyfin, Kodi, Emby, Nextcloud, PhotoPrism, and OBS Studio, are susceptible. Exploitation for remote code execution is feasible if Address Space Layout Randomization (ASLR) is disabled or bypassed. The widespread use of FFmpeg across various media applications amplifies the risk, highlighting the importance of prompt updates to mitigate potential attacks. This incident underscores the critical need for rigorous supply chain security practices and timely patch management to protect against emerging vulnerabilities.
1 month ago
Kill Chain
Global WhatsApp Phishing Campaign Exploits Fake Business Documents
In June 2026, a sophisticated phishing campaign targeted WhatsApp users globally, distributing malicious VBScript files disguised as business documents. Attackers compromised WhatsApp accounts to send these deceptive messages, leading recipients to execute scripts that disabled User Account Control (UAC) protections and installed ManageEngine Endpoint Central, granting remote access to victims' systems. The campaign affected users in countries including Brazil, India, Mexico, Singapore, the UK, Spain, Taiwan, Australia, Russia, Vietnam, and Malaysia. This incident underscores the evolving tactics of cybercriminals leveraging trusted communication platforms to disseminate malware. The use of legitimate software for malicious purposes highlights the need for heightened vigilance and robust security measures to protect against such sophisticated attacks.
1 month ago
Kill Chain
ShapedPlugin WordPress Pro Plugins Compromised in Supply Chain Attack
In June 2026, ShapedPlugin, a developer of premium WordPress plugins, experienced a supply chain attack where attackers compromised the company's update infrastructure. This breach led to the distribution of backdoored versions of several plugins, including Product Slider Pro for WooCommerce, Real Testimonials Pro, and Smart Post Show Pro. The malicious code, activated upon administrator access to the WordPress dashboard, connected to a command-and-control server to download additional payloads, resulting in unauthorized access and data exfiltration. ([thaicert.or.th](https://www.thaicert.or.th/en/2026/06/19/supply-chain-attack-through-shapedplugin-update-system-impacts-wordpress-websites/?utm_source=openai)) This incident underscores the growing threat of supply chain attacks targeting trusted software vendors. It highlights the critical need for organizations to implement robust security measures, including regular code audits and monitoring of update channels, to prevent similar compromises.
1 month ago
Kill Chain
DifyTap Vulnerabilities: A Wake-Up Call for Multi-Tenant AI Security
In June 2026, cybersecurity researchers from Zafran Security disclosed four critical vulnerabilities in Dify, an open-source agentic workflow platform. These flaws, collectively termed 'DifyTap,' allowed unauthorized access to AI conversations across different tenants without authentication. The vulnerabilities included authorization bypasses and path traversal issues, enabling attackers to read private AI chats, manipulate internal APIs, and access files across tenants. This incident underscores the growing risks associated with multi-tenant cloud services and the importance of stringent access controls. As AI platforms become integral to business operations, ensuring their security is paramount to prevent data breaches and maintain user trust.
1 month ago
Kill Chain
Crypto Heist Leveraging Fake Reputation Networks to Distribute Malware
In June 2026, cybercriminals orchestrated a sophisticated campaign to distribute a Rust-based clipboard hijacking malware targeting both Windows and macOS users. The attackers created a comprehensive fake reputation network, utilizing GitHub repositories, SourceForge projects, AI-generated YouTube videos, and manipulated VirusTotal comments to lend credibility to their malicious tools. These tools, masquerading as crypto trading and gambling aids, were designed to steal cryptocurrency by intercepting wallet addresses copied to the clipboard, affecting assets like Bitcoin, Ethereum, Monero, Binance Chain, and Solana. This incident underscores a significant evolution in cybercriminal tactics, highlighting their ability to exploit multiple trusted platforms to build false credibility and deceive users. The campaign's success demonstrates the urgent need for enhanced vigilance and skepticism towards online reputation signals, especially in the cryptocurrency domain, where the allure of quick profits can cloud judgment.
1 month ago
Kill Chain
React2Shell (CVE-2025-55182) Exploitation: A December 2025 Cybersecurity Incident
In December 2025, a critical vulnerability known as React2Shell (CVE-2025-55182) was disclosed, affecting React Server Components versions 19.0.0 through 19.2.0. This flaw allowed unauthenticated remote code execution via improper deserialization in the Flight protocol. Within hours of disclosure, multiple state-sponsored threat groups, including China's Earth Lamia and Jackpot Panda, as well as North Korean actors, began exploiting the vulnerability to deploy malware, establish persistent access, and exfiltrate data. The rapid exploitation led to significant security incidents across various sectors globally. ([aws.amazon.com](https://aws.amazon.com/blogs/security/china-nexus-cyber-threat-groups-rapidly-exploit-react2shell-vulnerability-cve-2025-55182/?utm_source=openai)) The React2Shell incident underscores the critical importance of prompt patching and vigilant monitoring. The swift exploitation by sophisticated threat actors highlights the need for organizations to enhance their vulnerability management processes and adopt proactive security measures to mitigate emerging threats effectively.
1 month ago
Kill Chain
Stop Active Cloud Data Exfiltration
Aviatrix Breach Lock helps teams instantly identify what data is leaving the environment, from which workload, and where it’s going — during an active breach.
Looking for threats in a different sector?
Browse All Threat Reports