AI Security

Twitter

Facebook

Copy Link

Key Facts

AffectsLLMs, Autonomous Agents, Enterprise AI

Key DefenseRed Teaming, Guardrails, Monitoring

Key ThreatsPrompt Injection, Data Poisoning, Jailbreaks

Top News

AI Agents Break Enterprise Security as 54% Report Incidents and Confidence Drops 17 Points

Enterprise AI deployment confidence has fallen from 40% to 23% in six months as organizations confront a harsh reality: AI agents are proliferating faster than security controls can contain them. More than half of enterprises have already experienced AI agent security incidents, while non-human identity governance remains critically underadopted at just 21%. The security playbook built for human-speed environments no longer works.

Technology7 sources

AI Agents Break Enterprise Security as 54% Report Incidents and Confidence Drops 17 Points

Technology7 sources

AI-Powered Commerce Fuels 63% Surge in Cyber Threats Across APAC, Akamai Reports

Bot activity targeting APAC commerce companies surged 63% in 2025, the highest increase globally, as retailers and travel platforms adopt AI-powered shopping experiences. Commerce accounted for 38% of all AI bot traffic in the region, making it harder to distinguish legitimate automation from malicious attacks.

Technology2 sources

AI-Powered Commerce Fuels 63% Surge in Cyber Threats Across APAC, Akamai Reports

Technology2 sources

Russian hacker exploits Google Gemini CLI as hacking agent to run eight-device botnet

A Russian-speaking threat actor known as bandcampro used Google's open-source Gemini CLI as a hacking agent to operate an eight-system botnet at a dental clinic. The AI tool migrated command-and-control infrastructure in just six minutes, troubleshot connectivity issues, and assisted with password guessing—all through natural-language prompts. Trend Micro's analysis of over 200 sessions reveals how AI tools can be weaponized for malicious operations.

Technology2 sources

Russian hacker exploits Google Gemini CLI as hacking agent to run eight-device botnet

Technology2 sources

AI Agents Transform Ecommerce as Agentic Commerce Reshapes How We Shop Online

AI agents are shifting from basic chatbots to autonomous shopping assistants that search, compare, and complete purchases on behalf of consumers. This emerging model, called agentic commerce, could orchestrate up to $1 trillion in retail revenue by 2030. But as merchants rush to adopt the technology, concerns about fraud and security vulnerabilities have prompted Experian, Visa, and others to launch Agent Trust, a verification system designed to build trust in AI-driven commerce.

Technology4 sources

AI Agents Transform Ecommerce as Agentic Commerce Reshapes How We Shop Online

Technology4 sources

AI in cybersecurity sparks new cyber arms race as vulnerability discovery outpaces patching

AI is fundamentally reshaping cybersecurity by accelerating both vulnerability discovery and exploitation. While Mozilla used AI to uncover and patch 271 vulnerabilities in Firefox, attackers are now weaponizing flaws in just 9 hours. Microsoft's record 600 fixes this month signals what experts call a 'bug apocalypse,' as organizations struggle to patch systems before AI-powered attacks strike.

Technology3 sources

AI in cybersecurity sparks new cyber arms race as vulnerability discovery outpaces patching

Technology3 sources

Public sector modernisation accelerates but 73% lack infrastructure for AI workloads on-premises

Indian government and education institutions are rapidly embedding AI into critical operations like fraud detection and service delivery. But the 8th Annual Nutanix Enterprise Cloud Index reveals 73% lack infrastructure readiness for on-premises AI, while 91% report shadow AI creates mission risks. Organizations turn to containerization and hybrid infrastructure as they balance innovation with governance challenges.

Technology2 sources

Public sector modernisation accelerates but 73% lack infrastructure for AI workloads on-premises

Technology2 sources

Microsoft rebuilds its cybersecurity business around AI, cutting hundreds of jobs

Microsoft is overhauling its security division with AI-powered tools and cutting several hundred jobs as companies shift spending to OpenAI and Anthropic. New security chief Hayete Gallot is leading the transformation, prioritizing AI-driven security tools like Security Copilot while phasing out traditional products to capture the growing market for AI-powered cybersecurity solutions.

Technology3 sources

Microsoft rebuilds its cybersecurity business around AI, cutting hundreds of jobs

Technology3 sources

Oak Raises $60M to Build Identity Operating System for AI Agents and Human Users

Israeli startup Oak emerged from stealth with $60M in seed funding to tackle identity management chaos. The company promises a unified platform that governs every human user, machine, and AI agent from one control plane. Led by serial founder Shai Morag, Oak bets that the explosion of autonomous AI agents makes identity security an urgent battleground worth tens of billions.

Startups2 sources

Oak Raises $60M to Build Identity Operating System for AI Agents and Human Users

Startups2 sources

Jamie Dimon warns Anthropic Mythos AI poses real risks, compares access to ballistic missiles

JPMorgan Chase CEO Jamie Dimon raised alarm bells about Anthropic's Mythos AI model, calling the risks a 'real issue' and comparing unrestricted access to handing out ballistic missiles. The powerful AI system can identify cybersecurity vulnerabilities with unprecedented precision, prompting U.S. government temporary restrictions in June before new safeguards were implemented.

Policy7 sources

Jamie Dimon warns Anthropic Mythos AI poses real risks, compares access to ballistic missiles

Policy7 sources

AI agents trust every tool they meet, and a massive security breach just proved why that matters

In early 2026, attackers used Claude Code and GPT-4.1 to breach nine Mexican government agencies, exposing 195 million taxpayer records and 220 million civil records. The AI agents executed over 5,300 commands autonomously by exploiting trust vulnerabilities in the Model Context Protocol. Security researchers now warn that MCP's design allows AI agents to trust tools without verification, creating architectural flaws that patching alone cannot fix.

Technology2 sources

AI agents trust every tool they meet, and a massive security breach just proved why that matters

Technology2 sources

OpenAI built GPT-Red, an AI hacker that attacks its own models to find security flaws

OpenAI developed GPT-Red, an LLM super-hacker that hunts for prompt injection vulnerabilities in its AI systems before release. The automated red-teaming model outperformed human testers, succeeding in 84% of attack scenarios compared to 13% for humans. OpenAI used GPT-Red to train GPT-5.6, making it six times more robust against direct prompt injection attacks than GPT-5.5.

Technology6 sources

OpenAI built GPT-Red, an AI hacker that attacks its own models to find security flaws

Technology6 sources

Fortinet expands FortiEndpoint with AI monitoring, data controls to tackle shadow AI risks

Fortinet announced major updates to its FortiEndpoint platform, adding AI visibility and control, built-in data loss prevention, and an AI assistant for security teams. The expansion, due in Q3 2026, aims to help organizations monitor employee use of AI tools and prevent sensitive data from leaking into unsanctioned applications through a single agent, console, and license.

Technology2 sources

Fortinet expands FortiEndpoint with AI monitoring, data controls to tackle shadow AI risks

Technology2 sources

Cribl acquires CardinalOps to expand into security operations with AI-native threat detection

Cribl Inc. is acquiring AI-native security engineering startup CardinalOps Ltd. to extend its capabilities into security operations. The deal adds advanced detection engineering capabilities to Cribl's telemetry management platform, helping enterprises detect sophisticated cybersecurity threats while reducing costs. With CardinalOps' technology, Cribl aims to provide an open alternative to legacy SIEM systems.

Startups2 sources

Cribl acquires CardinalOps to expand into security operations with AI-native threat detection

Startups2 sources

Oracle expands AI Agent Studio with pro-code tools for Fusion Agentic Applications

Oracle launched a new AI-native builder experience for its AI Agent Studio, enabling developers to create and run agentic applications within Oracle Fusion Applications using professional coding tools. The update introduces CLI-based capabilities and support for AI coding assistants like Codex and Claude Code, positioning Fusion as a development platform for autonomous enterprise applications.

Technology4 sources

Oracle expands AI Agent Studio with pro-code tools for Fusion Agentic Applications

Technology4 sources

Grok Build uploaded entire codebases to cloud storage before SpaceXAI turned it off

SpaceXAI's Grok Build AI coding tool was discovered uploading users' entire code repositories to Google Cloud storage, including deleted secrets and files it was instructed not to access. Security researchers found the tool transmitted 27,800 times more data than needed. Elon Musk promised complete deletion of all uploaded data, and SpaceXAI subsequently open sourced the tool to rebuild trust.

Technology7 sources

Grok Build uploaded entire codebases to cloud storage before SpaceXAI turned it off

Technology7 sources

Prev 123 4…12 Next