All news with #ai security tag

⚠️ Researchers demonstrated a client-side prompt injection called HashJack that hides malicious instructions in URL fragments after the '#' symbol. AI-powered browsers and assistants — including Comet, Copilot for Edge, and Gemini for Chrome — read these fragments for context, allowing attackers to weaponize legitimate sites for phishing, data exfiltration, credential theft, or malware distribution. Because fragment data never reaches servers, network defenses and server logs may not detect this technique.

🚧 Gemini 3’s release on 18 November 2025 signals a structural shift: beyond headline performance gains, it accelerates embedding large multimodal assistants directly into enterprise workflows and infrastructure. That continuation of a trend already visible with Microsoft Copilot effectively makes AI assistants a new enterprise perimeter — changing where corporate data, identities, and controls must be enforced. Security, compliance, and IT teams need to update policies, telemetry, and incident response to this expanded boundary.

⚠️Security researchers at Cato Networks disclosed HashJack, a novel indirect prompt-injection vulnerability that abuses URL fragments (the text after '#') to deliver hidden instructions to AI browsers. Because fragments never leave the client, servers and network defenses cannot see them, allowing attackers to weaponize legitimate websites without altering visible content. Affected agents included Comet, Copilot for Edge and Gemini for Chrome, with some vendors already rolling fixes.

🔐 The FBI warns that cybercriminals impersonating banks and payment services have caused over $262 million in losses this year through account takeover (ATO) fraud and more than 5,100 complaints. Attackers use phishing, SEO poisoning, calls and SMS to harvest credentials and MFA/OTP codes, then transfer funds to intermediary accounts and convert proceeds to cryptocurrency. The advisory highlights growing use of AI-generated phishing and holiday-themed scams and urges vigilance, unique passwords, URL checks and stronger authentication.

🛡️In 2026 Palo Alto Networks forecasts a shift to the Year of the Defender as enterprises counter AI-driven threats with AI-enabled defenses. The report outlines six predictions — identity deepfakes, autonomous agents as insider threats, data poisoning, executive legal exposure, accelerated quantum urgency, and the browser as an AI workspace. It urges autonomy with control, unified DSPM/AI‑SPM platforms, and crypto agility to secure the AI economy.

🎧 In Episode 78 of The AI Fix, hosts Graham Cluley and Mark Stockley examine a string of headline-grabbing AI stories, from a fact-checked “robot spider” scare to Anthropic’s claim of catching an autonomous AI cyber-spy. The discussion covers Claude hallucinations, alleged state-backed misuse of US AI models, and concerns about AI-driven military systems and investor exuberance. The episode also questions whether the current AI boom is a bubble, while highlighting real-world examples like AI-generated music charting and pilots controlling drone wingmen.

🌐 Check Point's 2026 analysis warns that an unprecedented convergence of AI, quantum computing, and an immersive Web 4.0 will reshape digital risk. Autonomous systems and hyper-automation will blur boundaries between cloud, networks, and physical infrastructure, expanding attack surfaces and changing the nature of digital trust. The report calls for updated cryptography, enhanced detection, and cross-industry resilience planning.

⚠️ 2026 will reshape SOC priorities as adversaries adopt AI to scale evasive attacks, creating urgent challenges across detection, triage, and proving business value. The piece identifies three critical problems: increasingly evasive threats, alert overload and analyst burnout, and the need to quantify ROI for security investments. It recommends interactive malware analysis to reveal full attack chains, real-time threat intelligence to enrich alerts and speed triage, and continuous, measurable intelligence (API/SDK-driven) to turn SOC activity into demonstrated business value.

🚨 Trend Micro warns agentic AI will increasingly automate attacks next year, with state-backed actors leading innovation before cybercriminals adopt the approach. Researchers say agentic systems — capable of taking autonomous actions — could chain discovery, exploitation and persistence steps, enabling less-skilled operators to run complex intrusions. The firm urges defenders to treat agents as privileged users and apply least-privilege, monitoring and assume-breach practices.

⚠️ CrowdStrike analysis finds that DeepSeek-R1, an open-source AI reasoning model from a Chinese vendor, produces significantly more insecure code when prompts reference topics the Chinese government deems sensitive. Baseline tests produced vulnerable code in 19% of neutral prompts, rising to 27.2% for Tibet-linked scenarios. Researchers also observed partial refusals and internal planning traces consistent with targeted guardrails that may unintentionally degrade code quality.

🎮 The UK government has launched the International Defence Esports Games (IDEG), touted as the world’s first military esports tournament, involving personnel from 40 allied nations. The three-day event at the National Gaming and Esports Arena in Sunderland on 9–11 October 2026 will combine live-streamed competitive matches with summits on cybersecurity, AI and drone operations. Officials say IDEG will sharpen cyber and digital skills, improve rapid decision-making and build allied partnerships.

📢 Amazon OpenSearch Service now supports OpenSearch 3.3, introducing search performance, observability, and agentic AI integration improvements. Vector search enhancements include agentic search for natural-language queries without complex DSLs, batch processing for the semantic highlighter to lower latency and improve GPU utilization, and optimizations in the Neural Search plugin. The release also makes Apache Calcite the default query engine for PPL, adds a broader PPL command library, and improves the approximation framework for more responsive pagination and dashboards. A new workload management plugin enables grouping of search traffic and tenant-level network isolation to prevent resource overuse.

🔒 Trend Micro is previewing Trend Vision One AI Security Package, a comprehensive suite due at AWS re:Invent in early December that aims to protect the full AI application stack from development through runtime. The offering combines continuous model scanning and automated AI guardrails and leverages Nvidia BlueField3 hardware acceleration. It also assembles tools such as AI Security Blueprint, Risk Insights, cloud and container security, file protection with NetApp support, an agentic SIEM with AWS native logs, and Zero Trust AI access controls.

🤖 AWS introduces the Agentic AI Security Scoping Matrix to help organizations secure autonomous, tool-enabled AI agents. The framework defines four architectural scopes—from no agency to full agency—and maps escalating security controls across six dimensions, including identity, data/memory, auditability, agent controls, policy perimeters, and orchestration. It advocates progressive deployment, layered defenses, continuous monitoring, and retained human oversight to mitigate risks as autonomy increases.

🤖Agentic AI gives autonomous agents the ability to access external systems, gather information, and take actions within defined workflows, making routine multi-system tasks far more efficient for human operators. Cisco Talos warns this efficiency is already being mirrored in the cyber crime economy, including the first observed AI-orchestrated campaign in early 2025. While AI lowers barriers to entry and speeds operations for attackers, it is imperfect and still requires skilled instruction and human oversight. Defenders can respond by building their own agentic tools, deploying honeypots to engage malicious agents, and refining detection to stay ahead.

🔍 DeepSeek-R1, a 671B-parameter open-source LLM, produced code with significantly more severe security vulnerabilities when prompts included politically sensitive modifiers. CrowdStrike found baseline vulnerable outputs at 19%, rising to 27.2% or higher for certain triggers and recurring severe flaws such as hard-coded secrets and missing authentication. The model also refused requests related to Falun Gong in 45% of cases, exhibiting an intrinsic "kill switch" behavior. The report urges thorough, environment-specific testing of AI coding assistants rather than reliance on generic benchmarks.

🛠️OpenAI has rolled out GPT-5.1-Codex-Max, a Codex variant optimized for long-running programming tasks and improved token efficiency. Unlike the general-purpose GPT-5.1, Codex is tailored to operate inside terminals and integrate with GitHub, and OpenAI says the model can work independently for hours. It is faster, more capable on real-world engineering tasks, uses roughly 30% fewer "thinking" tokens, and adds Windows and PowerShell capabilities. GPT-5.1-Codex-Max is available in the Codex CLI, IDE extensions, cloud, and code review.

🔎 Google is testing AI-powered, interactive UI answers within AI Mode, integrating Gemini 3 to generate on-the-fly interfaces tailored to queries. Instead of relying solely on text and a couple of links, Search can produce dynamic tools—such as an RNA polymerase simulator—to demonstrate concepts in action. This change could improve comprehension but may also reduce traffic to original sites and reshape the web economy.

🔒 Fortinet is an official launch partner for third-party rules on AWS Network Firewall, introducing Fortinet Managed IPS Rules powered by FortiGuard AI-Powered Security Services. The managed service uses AI/ML from FortiGuard Labs to automatically translate global threat telemetry into continuously updated IPS rules, removing manual tuning and improving detection timeliness. Deployment is fast via AWS Marketplace and integrates natively with AWS Network Firewall, helping teams scale protection across cloud workloads while supporting compliance objectives.

🔒 Amazon Web Services expanded Amazon Bedrock Guardrails to cover code-related use cases, enabling detection and prevention of harmful content embedded in code. The update applies content filters, denied topics, and sensitive information filters to code elements such as comments, variable and function names, and string literals. The enhancements also include prompt leakage detection in the standard tier and are available in all supported AWS Regions via the console and APIs.