All news with #prompt injection attack tag

🔍 New research finds current AI web agents largely fail to defend against prompt injection attacks. The StakeBench benchmark tested GPT‑5 and Gemini‑powered agents across realistic web scenarios, revealing high success rates for both direct and indirect injections and exposing failure modes like stealthy parasitism and misaligned disruption. Results show vulnerabilities vary by stakeholder and agent architecture.

🛡️ Two security teams demonstrated attacks against OpenClaw, where hidden instructions in shared contacts, vCards, and location pins or ordinary-looking emails caused the agent to execute attacker-controlled code or exfiltrate sensitive data. Imperva found a message-object prompt-injection flaw that OpenClaw patched in version 2026.4.23, while Varonis showed social-engineering 'agent phishing' that requires architectural controls rather than a simple patch. Operators are urged to update, restrict outbound actions, and treat agents as junior employees needing human oversight.

🛡️ An enterprise sales team built an AI agent to manage renewals; the agent reads emails, queries CRM data, drafts responses, and updates records. This workflow combines private data, untrusted input, and external communication, changing the security model. Traditional controls like IAM and DLP still matter but are insufficient alone. Runtime, context-aware controls that inspect prompts, outputs, and tool calls are required to prevent prompt injection, data exfiltration, and unsafe actions.

🔒 OpenAI has rolled out two new security controls for ChatGPT: Lockdown Mode and Active Sessions. Lockdown Mode restricts outbound network access to prevent data exfiltration via prompt injection, at the cost of disabling live connectors and certain features. Active Sessions gives users visibility into and control over signed-in devices, with the ability to end single or all sessions. Both controls target account security and sensitive-data use cases, though SSO accounts and some logins remain unsupported.

🛡️ Ariel Fogel warned at Infosecurity Europe 2026 that prompt injection is an unresolved architectural issue threatening AI development. He explained that LLMs treat inputs as a single token stream, preventing reliable privilege separation between system prompts, user inputs and agent-retrieved content. With agents gaining tool access, successful injections can escalate from bad outputs to real-world actions, outpacing traditional governance and controls.

🔒 OpenAI has started rolling out a new Lockdown Mode for eligible ChatGPT personal accounts to reduce the risk of data exfiltration from prompt injection attacks. The optional security setting restricts capabilities that can connect to the web or external services, including live web browsing, image support, agent mode, deep research, Canvas networking, and file downloads. Lockdown Mode is available across Free, Go, Plus, Pro, and self-serve ChatGPT Business plans but cannot be used simultaneously with Developer Mode. OpenAI warns the feature reduces but does not eliminate exfiltration risk and also launched enhanced account session management to help detect and terminate unauthorized access.

🔒 A researcher discovered a vulnerability in Anthropic's Claude Code GitHub Action that allowed takeover of public repositories via a single opened GitHub issue. Anthropic patched the core bypass in January and released fixes in claude-code-action v1.0.94, rating the issue 7.8 under CVSS v4.0 and issuing a bounty. The flaw arose from overly permissive triggers that trusted actors ending in "[bot]" and example workflows allowing non-write users, enabling indirect prompt injection to exfiltrate environment secrets and OIDC credentials. Administrators should update to v1.0.94, audit workflows for untrusted inputs, and remove unnecessary permissions and tools to prevent exfiltration.

🔔 A SafeBreach researcher demonstrated that a single malicious notification from apps like WhatsApp, Slack, SMS, Signal, Instagram, or Messenger could hijack Google Gemini's voice assistant on Android. The technique, called Fake Context Alignment, let notifications be treated as executable context, enabling fake replies, app launches, smart-home control, and even persistent memory poisoning. Google patched the issue server-side after being notified; Android users can disable Gemini's notification reading to mitigate exposure.

🔒 Reporting suggests attackers used Meta’s AI support chatbot to change recovery emails on high-profile Instagram accounts, leading to notable takeovers. The core issue isn’t just prompt injection or a model jailbreak but that the AI operated within a sensitive account recovery workflow with insufficient independent verification. Organizations must treat AI-driven support actions as part of the security boundary and constrain authority, permissions, and verification around such agents.

🛡️ Cybersecurity researchers disclosed a vulnerability dubbed ChatGPhish that exploits ChatGPT's trust in Markdown links and images to perform prompt injections and enable phishing. The flaw causes the assistant to auto-fetch attacker-hosted images and render malicious links and QR codes inside the trusted UI, potentially leaking client metadata like IP and User-Agent. The technique highlights summarization as an adversarial surface that can convert benign web pages into phishing vectors.

🔍 Cisco researchers found that popular frontier LLMs from OpenAI, Anthropic, Google, xAI, and Amazon exhibit substantially higher risk when subjected to multi-turn adversarial attacks than when assessed with single-prompt safety benchmarks. The team ran tens of thousands of single-turn and multi-turn attacks across 15 models and multiple configurations, revealing wide gaps in attack success rates (ASRs) and configuration-dependent safety behavior. They urge improved benchmarks, transparency on configuration impacts, and publication of paired single- and multi-turn ASRs to better inform procurement and governance decisions.

🛡️ Check Point explains why GenAI chatbots create new security risks by acting as a front door to internal systems and data. The post highlights real incidents—prompt injection, data exposure, and misleading responses—that demonstrate legal, financial, and reputational impacts. It describes how Check Point WAF extends unified application and API security into the conversational layer to detect and block malicious prompts, prevent data leaks, and control unsafe outputs.

🔍 Researchers from Xidian University describe a new image-based prompt injection called CrossMPI that uses near-imperceptible pixel perturbations to alter how large vision-language models interpret both visual and textual inputs. The technique targets intermediate multimodal fusion layers rather than final outputs, misleading LVLMs without modifying text prompts. Tests show strong black-box transferability and high success rates across several open-source models, while common defenses reduce but do not fully eliminate the threat.

🔒 Penetration testing shows AI and LLM deployments contain a disproportionate share of severe vulnerabilities. Cobalt’s State of Pentesting Report finds 32% of LLM findings rated high risk versus 13% for legacy enterprise tests, and only 38% of those high-risk LLM issues are remediated. Experts point to emerging attack surfaces — notably prompt injection, now OWASP’s top LLM risk — broader blast radii from model integrations, and fragmented ownership for fixes. Recommended countermeasures include threat modeling, red teaming, least-privilege access, strict output validation, and human approval gates for high-consequence actions.

⚠️ Microsoft researchers disclosed critical vulnerabilities in Semantic Kernel that allow prompt injection to escalate into host-level remote code execution and arbitrary file writes. The team detailed two fixed issues — CVE-2026-26030 (unsafe eval-style filter in the In-Memory Vector Store) and CVE-2026-25592 (exposed DownloadFileAsync in SessionsPythonPlugin) — and provided mitigations. Operators should upgrade the Python package to 1.39.4+ and the .NET SDK to 1.71.0+, validate any model-influenced tool parameters as untrusted input, and hunt endpoint telemetry for post-exploitation indicators.

⚠️ ReversingLabs researchers describe an ongoing supply‑chain campaign called PromptMink that manipulates AI coding agents into installing malicious dependencies. Attackers publish bait packages with persuasive READMEs and LLM‑optimized documentation on registries like NPM and PyPI to increase discovery by autonomous agents and developers. The operation, attributed to North Korea’s Famous Chollima, paired legitimate‑looking SDKs with second‑layer packages carrying infostealers, later evolving to compiled Rust add‑ons, SEAs, SSH backdoors, and project exfiltration.

🔔 LayerZero-linked infrastructure poisoning likely enabled a North Korean-linked group (TraderTraitor/TraderTraiter) to steal $290M from KelpDAO by compromising RPC nodes and exploiting a quorum while a DDoS distracted a third node, prompting an Arbitrum Security Council freeze. At the same time, active RCE attacks, malicious npm packages delivering credential stealers and SSH backdoors, and indirect AI prompt injection payloads are accelerating breaches. The bulletin also flags covert browser access by desktop AI apps, a surge in commodified malware, SIM-farm services, and persistent exploitation of long-known weaknesses; the practical remedies remain patch early, verify dependencies, and restrict implicit trust.

🛡️ Google unveiled a suite of agentic AI defenses at Google Cloud Next '26 to help SOC teams manage a surge of vulnerabilities tied to Anthropic Mythos. The launch includes three new agents in Google Security Operations — threat hunting, detection engineering, and third-party context — plus expanded Wiz integrations and an AI-BOM to inventory AI components. Additional controls like Agent Identity, Agent Gateway, and Model Armor aim to govern the emerging 'agentic web' and mitigate prompt injection, data leakage, and shadow AI risks.

⚠️ Google’s Antigravity IDE contained a prompt-injection flaw that could convert a file-search operation into remote code execution. Researchers at Pillar Security showed the agent’s find_my_name tool passed unsanitized Pattern strings to the underlying fd utility, allowing flag injection and execution of binaries. Google acknowledged and fixed the issue and awarded a VRP bounty, but the flaw underscores limits of shell-focused sanitization.

🛡️ Google has patched a critical prompt-injection vulnerability in its agentic IDE Antigravity that could allow attackers to achieve arbitrary code execution. Researchers at Pillar Security found that the find_by_name tool passed unsanitized input to the native fd search utility, enabling injection of the -X (exec-batch) flag to run staged scripts. Because this call executes before Strict Mode constraints are applied, an attacker can stage a malicious file and trigger it via a crafted search pattern. The issue was disclosed January 7 and fixed by Google on February 28.