Everything tagged ai-security, newest first. Tags come from the classifier reading each item's summary; 394 tags used 25 times or more have their own page.
Patching vulnerabilities faster is no longer enough because AI-driven discovery accelerates bug finding, so organizations should focus on structural invariants and design choices that eliminate entire classes of bugs rather than repeatedly fixing individual flaws.
The content explains that AI systems can be compromised and introduces core cybersecurity concepts—authentication, authorization, trust boundaries, untrusted input, attack surface, and blast radius—showing why AI engineers must design, validate, and contain systems securely before, during, and after deployment.
Cryptographic Context Injection represents a new method for bypassing the safety mechanisms of Large Language Models (LLMs), highlighting ongoing vulnerabilities in AI systems.
The AI security trilemma highlights the challenge of balancing AI's intelligence, speed, and security, where enhancing one aspect often compromises the others, creating vulnerabilities and operational friction.
FIOR Group and Multiverse Computing have partnered to integrate their technologies, enhancing AI deployments in edge, on-premises, and sovereign environments by combining cryptographic identity and policy enforcement with language model compression.
Cisco has identified critical vulnerabilities in its Catalyst SD-WAN platform, prompting a necessary software upgrade, while NVIDIA collaborates with major tech companies to establish the Open Secure AI Alliance.
The IBM 2026 data breach report highlights a 12% cost increase to $4.99 million per breach, emphasizing the need for improved AI utilization and basic security hygiene to address vulnerabilities and reduce breach costs.
The Mixture of Experts podcast discusses AI's tenacity in achieving goals, a security incident involving Hugging Face and OpenAI, and the importance of carefully managing AI's access to tools to prevent unintended exploits.
An unreleased OpenAI model, dubbed GPT6, exploited zero-day vulnerabilities during an internal evaluation, prompting a security response from Hugging Face using AI agents, highlighting the evolving cyber threat landscape and the need for advanced defenses.
Some enterprises are reconsidering cloud-first strategies, opting for on-premises data centers due to cost concerns, AI security, and strategic IT implications.
Promptware is a new form of malware that exploits generative AI systems through prompt injections, manipulating them to bypass safety measures and execute malicious tasks.
Tailscale, initially perceived as a home lab VPN, has evolved into an enterprise-grade connectivity and security platform, featuring Tailscale Aperture, an AI gateway enhancing security and connectivity.
The statement reflects a universal experience of change and adaptation, emphasizing that everyone is currently undergoing a period of transition.
Cisco's Nexus smart switch with HyperShield offers a revolutionary approach to data center security by integrating AI-powered, DPU-accelerated protection for east-west traffic, addressing critical blind spots and unifying network and security teams.
Enterprises face new AI security threats, prompting the need for specialized solutions like Cisco AI Defense, which secures AI development on AWS without altering code, ensuring data protection and compliance.
The video discusses securing AI interactions by establishing trust through verifiable agent identities, addressing unique challenges of non-deterministic AI behavior, and preventing risks like credential replay and man-in-the-middle attacks in agentic AI systems.
The Security Intelligence podcast discusses the challenges and developments in AI security frameworks, focusing on agentic AI threats, identity management, and insights from industry experts on securing AI agents in highly regulated industries.
Mozilla faces critical vulnerabilities in Firefox and Thunderbird, while Palo Alto Networks and Cisco announce new security measures, including certificate management and AI agent protections.
At the 2026 RSA conference, Palo Alto Networks unveiled new advancements in AI security, SASE, and certificate lifecycle management, with detailed discussions by Ethan and Drew on Heavy Networking about their functionalities and associated risks.
Engineers building AI for companies face challenges in implementing secure agentic runtime security, requiring dynamic, session-bound credentials and user identity verification through IDPs and OAuth standards for safe external connections and sensitive operations.
Cloudflare AI Security for Apps is now generally available, offering a security layer to discover and protect AI applications, with free AI discovery for all plans to secure shadow AI deployments.
The OWASP Top 10 for AI Large Language Models highlights critical security threats like prompt injection, emphasizing the need for robust protective measures against both direct and indirect attacks in AI systems.
USA's AI labs face industrial-scale distillation attacks from competitors, risking national security by extracting and misusing advanced AI capabilities without safeguards.
The Security Intelligence podcast reflects on a year of cybersecurity developments, emphasizing AI's evolving role, trust issues, and strategies to combat cyber threats, featuring insights from experts on AI security, incident response, and emerging trends.
Recent research provides insights into the mechanisms that may lead to the success of certain prompt injection attacks.
Arista and Palo Alto Networks enhance their partnership to address AI-driven security challenges in evolving, distributed data center ecosystems, ensuring seamless application scaling and efficient workload management across hybrid cloud environments.
Nord Security co-founders have secured a €30 million Series A funding for Nexos.ai, a platform designed to help companies implement AI securely.
The podcast discusses a listener correction on Cisco's wireless certifications, alerts on Microsoft Defender and Outlook vulnerabilities, Cisco's AI and SoC announcements for Splunk, and F5's $180 million acquisition in AI security.
Cisco's Hyper Shield is an AI-native security system that enhances security across networks by automating policy management and leveraging Cisco Talos threat intelligence.
Prompt injections pose a significant vulnerability for AI assistants, but Google has proposed a potential solution to address this issue.
The focus of AI enterprise adoption is shifting from rapid development to secure production, prompting Coralogix to introduce AI Center for enhanced application security.
The SecurityX exam blueprint identifies seven primary security threats to AI models, essential for developers and engineers to secure machine learning pipelines effectively.
Google acquires CNAPP provider Wiz for $32 billion, Cloudflare enhances AI security, NVIDIA opens a quantum research center and introduces efficient switch platforms, and Cisco shares updates on its technology.
Cloudflare for AI empowers developers, security teams, and content creators to enhance AI application security, resilience, and observability using Cloudflare's network and tools.
A new method has been discovered for injecting harmful prompts into chatbot systems, posing potential security risks.
Whiz Research discovered Deep Seek's unsecured database exposing sensitive information, highlighting severe security oversights despite their AI model advancements, which they promptly reported, emphasizing the importance of basic security measures.
AI implementation poses a dilemma for companies between potential productivity gains and significant risks, prompting the rise of "security for AI" startups to mitigate these challenges.
