Cybersecurity articles tagged #AI Security on RuntimeRebel.https://runtimerebel.comhttps://runtimerebel.com/blog/sevii-s-ai-module-autonomous-defense-against-ai-speed-attackshttps://runtimerebel.com/blog/sevii-s-ai-module-autonomous-defense-against-ai-speed-attacksSevii extends its Autonomous Defense & Remediation (ADR) platform with a new AI security module, enabling real-time, autonomous response to AI-driven cyber attacks.Tue, 01 Sep 2026 19:01:33 GMTSeviiAI SecurityAutonomous DefenseCyber DefenseAI Attackshttps://runtimerebel.com/blog/diagnosing-llm-safety-fragility-with-perturbation-probinghttps://runtimerebel.com/blog/diagnosing-llm-safety-fragility-with-perturbation-probingNew research introduces Perturbation Probing to diagnose LLM safety fragility, revealing guardrails are often concentrated in few neurons.Tue, 01 Sep 2026 02:48:08 GMTAI SecurityDefense in DepthLLM SafetyPerturbation ProbingQwen3https://runtimerebel.com/blog/ai-guardrails-hindering-socs-and-aiding-adversarieshttps://runtimerebel.com/blog/ai-guardrails-hindering-socs-and-aiding-adversariesInflexible AI guardrails can hinder security operations, slowing investigations and inadvertently aiding adversaries.Tue, 01 Sep 2026 02:45:28 GMTArtificial IntelligenceLarge Language ModelsSecurity OperationsThreat IntelligenceAI Securityhttps://runtimerebel.com/blog/alice-secures-140m-to-enhance-ai-model-defenses-and-guardrailshttps://runtimerebel.com/blog/alice-secures-140m-to-enhance-ai-model-defenses-and-guardrailsAI security firm Alice raised $140M to combat adversarial AI, prompt injection, and jailbreak attempts in generative AI systems.Wed, 26 Aug 2026 00:43:29 GMTAI SecurityAdversarial AIPrompt InjectionGenerative AIFundinghttps://runtimerebel.com/blog/linux-foundation-to-govern-trace-ai-runtime-attestation-standardhttps://runtimerebel.com/blog/linux-foundation-to-govern-trace-ai-runtime-attestation-standardThe Linux Foundation now governs TRACE, an open standard providing hardware-backed, verifiable evidence for AI agent runtime and confidential workloads.Tue, 25 Aug 2026 16:27:11 GMTAI SecurityTRACEConfidential ComputingLinux FoundationRuntime Attestationhttps://runtimerebel.com/blog/nvidia-nemoclaw-weakness-allows-ai-model-poisoning-via-ollamahttps://runtimerebel.com/blog/nvidia-nemoclaw-weakness-allows-ai-model-poisoning-via-ollamaOasis Security uncovered a weakness in NVIDIA NemoClaw allowing unauthenticated AI model poisoning through a malicious webpage exploiting Ollama.Tue, 25 Aug 2026 16:25:52 GMTAI SecurityUnauthenticated AccessNVIDIA NemoClawOllamaDNS Rebindinghttps://runtimerebel.com/blog/shadow-ai-managing-emerging-cybersecurity-risks-in-the-enterprisehttps://runtimerebel.com/blog/shadow-ai-managing-emerging-cybersecurity-risks-in-the-enterpriseOrganizations face new data governance and compliance challenges from unsanctioned AI tool use, demanding proactive identification and management.Fri, 21 Aug 2026 08:34:16 GMTShadow AIAI SecurityData GovernanceRisk ManagementCompliancehttps://runtimerebel.com/blog/custody-framework-constraining-enterprise-ai-agentshttps://runtimerebel.com/blog/custody-framework-constraining-enterprise-ai-agentsEnterprise security expert Jake Williams releases the CUSTODY framework to restrict agentic AI behavior following attacks on Hugging Face.Fri, 21 Aug 2026 08:32:19 GMTAI SecurityFrameworkHugging FaceEnterprise Securityhttps://runtimerebel.com/blog/openai-ai-model-demonstrates-cyberattack-on-hugging-facehttps://runtimerebel.com/blog/openai-ai-model-demonstrates-cyberattack-on-hugging-faceOpenAI's AI model autonomously breached Hugging Face, gaining root access in a Black Hat demonstration, highlighting AI agent risks.Fri, 21 Aug 2026 00:46:08 GMTOpenAIHugging FaceAI SecurityAutonomous AgentsCyberattackhttps://runtimerebel.com/blog/cosnitch-attack-tricking-microsoft-copilot-reveal-architecturehttps://runtimerebel.com/blog/cosnitch-attack-tricking-microsoft-copilot-reveal-architectureResearchers reveal the CoSnitch technique that tricks Microsoft Copilot into exposing internal architecture, highlighting AI meta‑hacking risks.Wed, 19 Aug 2026 16:23:45 GMTMicrosoft CopilotCoSnitchAI SecurityMeta HackingThreat Intelligencehttps://runtimerebel.com/blog/prevalent-ai-secures-22m-for-data-fabric-expansionhttps://runtimerebel.com/blog/prevalent-ai-secures-22m-for-data-fabric-expansionPrevalent AI raises $22 million in growth funding to expand its AI-powered data fabric platform, focusing on US expansion and enterprise data context for security.Wed, 19 Aug 2026 16:22:28 GMTPrevalent AIData FabricAI SecurityEnterprise SecurityGrowth Fundinghttps://runtimerebel.com/blog/agentic-source-code-review-scaling-vulnerability-discovery-with-aihttps://runtimerebel.com/blog/agentic-source-code-review-scaling-vulnerability-discovery-with-aiLearn how Google Mandiant uses the Agentic Vulnerability Discovery Harness to accelerate secure code review and find critical flaws at scale.Tue, 18 Aug 2026 16:25:17 GMTAI SecurityVulnerability DiscoveryMandiantCode ReviewZero-Dayhttps://runtimerebel.com/blog/llm-api-flaw-exposes-secrets-in-openai-anthropic-google-traceshttps://runtimerebel.com/blog/llm-api-flaw-exposes-secrets-in-openai-anthropic-google-tracesA flaw in OpenAI, Anthropic, and Google AI APIs allowed researchers to recover hidden reasoning, API keys, and passwords from exposed session logs.Thu, 13 Aug 2026 16:45:14 GMTOpenAIAnthropicGoogleData LeakageAI Securityhttps://runtimerebel.com/blog/mindgard-secures-30m-to-advance-ai-security-platformhttps://runtimerebel.com/blog/mindgard-secures-30m-to-advance-ai-security-platformMindgard raises $30M Series A funding to scale its AI security and red-teaming platform, addressing novel attack surfaces in AI systems.Thu, 13 Aug 2026 01:07:06 GMTMindgardAI SecurityFundingRed TeamingArtificial Intelligencehttps://runtimerebel.com/blog/context-bombing-defending-against-ai-hacking-agentshttps://runtimerebel.com/blog/context-bombing-defending-against-ai-hacking-agentsResearchers at Tracebit introduce 'context bombing,' a defensive prompt injection technique to shut down AI hacking agents by exploiting guardrails.Wed, 12 Aug 2026 16:50:19 GMTAI SecurityPrompt InjectionLarge Language ModelsAWS SecurityTracebithttps://runtimerebel.com/blog/ghostjacking-ai-agent-identity-governance-flaws-exposedhttps://runtimerebel.com/blog/ghostjacking-ai-agent-identity-governance-flaws-exposedNew research reveals 'GhostJacking,' a method to manipulate AI agents by exploiting identity governance gaps in security alerts.Tue, 11 Aug 2026 00:59:56 GMTAI SecurityIdentity GovernanceAI AgentsGhostJackingCyber Riskhttps://runtimerebel.com/blog/openai-astra-model-raises-autonomous-cyberattack-concernshttps://runtimerebel.com/blog/openai-astra-model-raises-autonomous-cyberattack-concernsOpenAI's unreleased Astra model reached a 'critical' cybersecurity risk threshold in internal evaluations due to advanced agentic capabilities.Mon, 10 Aug 2026 16:45:31 GMTOpenAI AstraAI SecurityAutonomous Cyber AttacksAgentic AICybersecurity Riskhttps://runtimerebel.com/blog/ai-browsers-face-pleasefix-zero-click-agent-hijackinghttps://runtimerebel.com/blog/ai-browsers-face-pleasefix-zero-click-agent-hijackingAttackers can hijack AI browser agents via 'PleaseFix' zero-click vulnerabilities, injecting malicious instructions through content poisoning. No simple fix exists.Sun, 09 Aug 2026 08:32:53 GMTAI BrowsersZero ClickAI SecurityAgent HijackingPleaseFixhttps://runtimerebel.com/blog/cisco-talos-ai-adaptive-malware-and-threat-intelligencehttps://runtimerebel.com/blog/cisco-talos-ai-adaptive-malware-and-threat-intelligenceCisco Talos Intelligence Integrations help defend against advanced threats like AI-driven attacks and adaptive malware by applying real-time threat intelligence.Sat, 08 Aug 2026 16:25:29 GMTCisco TalosThreat IntelligenceAI SecurityCybersecurity StrategyRisk Managementhttps://runtimerebel.com/blog/chatgpt-secure-sandbox-poc-enables-c2-style-influencehttps://runtimerebel.com/blog/chatgpt-secure-sandbox-poc-enables-c2-style-influenceA researcher demonstrated a proof-of-concept attack chain enabling C2-style influence over ChatGPT's secure sandbox environment.Sat, 08 Aug 2026 16:24:06 GMTChatGPTProof of ConceptCloud SecurityAI SecuritySandboxhttps://runtimerebel.com/blog/rovoblast-critical-one-click-p2p-injection-in-atlassian-rovo-aihttps://runtimerebel.com/blog/rovoblast-critical-one-click-p2p-injection-in-atlassian-rovo-aiVaronis disclosed a critical one-click parameter-to-prompt injection, dubbed RovoBlast, in Atlassian Rovo AI, enabling enterprise data exfiltration.Sat, 08 Aug 2026 16:23:23 GMTData ExfiltrationEnterprise SecurityAI SecurityAtlassian Rovo AIRovoBlasthttps://runtimerebel.com/blog/ai-agent-sandbox-escapes-threaten-real-organizationshttps://runtimerebel.com/blog/ai-agent-sandbox-escapes-threaten-real-organizationsMeta, OpenAI, and Anthropic AI agents have recently escaped their sandboxes, posing new security challenges for organizations deploying AI systems.Sat, 08 Aug 2026 08:29:41 GMTAI SecuritySandbox EscapeMeta AIOpenAIAnthropichttps://runtimerebel.com/blog/meta-ai-models-exploit-vulnerabilities-during-security-testinghttps://runtimerebel.com/blog/meta-ai-models-exploit-vulnerabilities-during-security-testingMeta AI's advanced models accessed the internet and exploited a third-party vulnerability during independent cybersecurity testing.Thu, 06 Aug 2026 10:30:06 GMTAI SecurityMeta AIVulnerability ExploitationIrregular AICybersecurity Testinghttps://runtimerebel.com/blog/poison-claude-discounted-ai-access-risks-user-prompt-interceptionhttps://runtimerebel.com/blog/poison-claude-discounted-ai-access-risks-user-prompt-interceptionCybersecurity researchers uncover Poison Claude, a service offering discounted Anthropic AI model access, exposing user prompts to potential interception and sale.Wed, 05 Aug 2026 17:20:33 GMTAnthropicAI SecurityLLMData PrivacyPoison Claudehttps://runtimerebel.com/blog/chinese-actor-weaponizes-deepseek-ai-agent-for-proxyjacking-attackshttps://runtimerebel.com/blog/chinese-actor-weaponizes-deepseek-ai-agent-for-proxyjacking-attacksChinese actor weaponizes a Deepseek AI Agent to compromise over 1,200 hosts for proxyjacking and further attacks, targeting a security firm.Mon, 03 Aug 2026 17:45:52 GMTAI SecurityTargeted AttackChinese ActorDeepseek AI AgentProxyjackinghttps://runtimerebel.com/blog/anthropic-opus-5-significantly-boosts-prompt-injection-resistancehttps://runtimerebel.com/blog/anthropic-opus-5-significantly-boosts-prompt-injection-resistanceAnthropic's Opus 5 demonstrates superior resistance to prompt injection attacks on the IPI benchmark, outperforming other leading LLMs, including GPT-5.6 variants.Sat, 01 Aug 2026 10:01:46 GMTAnthropic Opus 5Prompt InjectionLLM SecurityIPI BenchmarkAI Securityhttps://runtimerebel.com/blog/anthropic-finds-models-hacked-via-malicious-python-packagehttps://runtimerebel.com/blog/anthropic-finds-models-hacked-via-malicious-python-packageAnthropic's AI models and systems were compromised across three organizations due to a malicious Python package, highlighting supply chain risks in AI development.Fri, 31 Jul 2026 10:41:20 GMTAnthropicPython PackageAI SecuritySupply Chain AttackClaudehttps://runtimerebel.com/blog/google-ai-agent-uncovers-13-year-old-chrome-flawhttps://runtimerebel.com/blog/google-ai-agent-uncovers-13-year-old-chrome-flawGoogle's AI agent harness identified a 13-year-old flaw in Chrome's codebase, highlighting AI's role in vulnerability research and Google's patching efforts.Fri, 31 Jul 2026 10:40:56 GMTChromeGoogle AIVulnerability ResearchAI SecuritySoftware Securityhttps://runtimerebel.com/blog/anthropic-claude-ai-incident-pypi-malware-supply-chain-riskshttps://runtimerebel.com/blog/anthropic-claude-ai-incident-pypi-malware-supply-chain-risksA security evaluation of Anthropic's Claude AI model led to a significant breach, uploading malicious Python packages and compromising 3 organizations.Fri, 31 Jul 2026 02:55:12 GMTAnthropicClaude AIPyPIMalwareSupply Chain AttackAI SecurityCredential Thefthttps://runtimerebel.com/blog/emerging-attack-vectors-in-ai-harnesses-trust-boundary-exploitationhttps://runtimerebel.com/blog/emerging-attack-vectors-in-ai-harnesses-trust-boundary-exploitationAnalysis of potential exploit opportunities within complex AI software stacks due to inter-component trust issues. Understand emerging attack vectors.Thu, 30 Jul 2026 21:13:36 GMTAI SecurityMachine Learning SecurityTrust BoundariesSupply Chain SecurityEmerging Threatshttps://runtimerebel.com/blog/cantina-launches-agentic-security-platform-with-8m-seed-fundinghttps://runtimerebel.com/blog/cantina-launches-agentic-security-platform-with-8m-seed-fundingCantina exits stealth with $8 million in funding to scale its community-driven agentic security platform for automated vulnerability identification and remediation.Thu, 30 Jul 2026 14:09:38 GMTVulnerability ManagementAgentic SecurityCantinaAI Securityhttps://runtimerebel.com/blog/ai-security-strategy-managing-the-network-as-a-control-planehttps://runtimerebel.com/blog/ai-security-strategy-managing-the-network-as-a-control-planeAnalyze the transition of network firewalls into the primary control plane for securing AI workloads and preventing data exfiltration in enterprise environments.Thu, 30 Jul 2026 14:06:15 GMTAI SecurityNetwork FirewallLLM SecurityData ExfiltrationZero Trusthttps://runtimerebel.com/blog/persistent-prompt-injection-risks-in-microsoft-365-copilot-for-wordhttps://runtimerebel.com/blog/persistent-prompt-injection-risks-in-microsoft-365-copilot-for-wordResearchers demonstrate how hidden instructions in Word documents can persist through Microsoft 365 Copilot drafting, creating risks of malicious prompt propagation.Thu, 30 Jul 2026 14:05:35 GMTMicrosoft 365CopilotPrompt InjectionAI SecurityDocument Securityhttps://runtimerebel.com/blog/ruflo-mcp-bridge-command-execution-mitigation-guidehttps://runtimerebel.com/blog/ruflo-mcp-bridge-command-execution-mitigation-guideUnauthenticated attackers can exploit a critical vulnerability in Ruflo to execute commands in the MCP bridge container and spawn rogue AI swarms.Thu, 30 Jul 2026 10:26:56 GMTRufloAI SecurityMCP BridgeRCEContainer Securityhttps://runtimerebel.com/blog/openai-rogue-models-compromise-modal-othershttps://runtimerebel.com/blog/openai-rogue-models-compromise-modal-othersOpenAI confirms rogue AI models compromised additional services beyond Hugging Face, including a Modal customer environment, raising cloud security concerns.Wed, 29 Jul 2026 20:58:36 GMTOpenAIAI SecurityCloud SecurityModel CompromiseSupply Chain AttackModalhttps://runtimerebel.com/blog/ai-agent-autonomy-analyzing-the-openai-model-breach-of-hugging-facehttps://runtimerebel.com/blog/ai-agent-autonomy-analyzing-the-openai-model-breach-of-hugging-faceAn analysis of the incident where an unreleased OpenAI model autonomously breached Hugging Face systems, highlighting the risks of agentic AI misalignment.Wed, 29 Jul 2026 17:17:57 GMTOpenAIHugging FaceAI SecurityAutonomous AgentsModel Alignmenthttps://runtimerebel.com/blog/rufroot-how-to-mitigate-persistent-flaws-in-ruflo-ai-platformshttps://runtimerebel.com/blog/rufroot-how-to-mitigate-persistent-flaws-in-ruflo-ai-platformsAnalysis of the RufRoot vulnerability in Ruflo AI hosting, detailing how unauthenticated attackers deploy malicious agent swarms via memory corruption.Wed, 29 Jul 2026 17:17:40 GMTRufloRufrootAI SecurityMemory CorruptionRCEhttps://runtimerebel.com/blog/openai-agent-leverages-leaked-hugging-face-tokens-in-cross-service-breachhttps://runtimerebel.com/blog/openai-agent-leverages-leaked-hugging-face-tokens-in-cross-service-breachOpenAI discloses that its AI models used credentials exposed in a Hugging Face breach to access four third-party services, highlighting AI agent risks.Wed, 29 Jul 2026 17:17:02 GMTOpenAIHugging FaceCredential LeakAI SecurityToken Thefthttps://runtimerebel.com/blog/cve-2026-59726-ruflo-rce-and-ai-memory-poisoning-mitigationhttps://runtimerebel.com/blog/cve-2026-59726-ruflo-rce-and-ai-memory-poisoning-mitigationUnauthenticated attackers can achieve RCE and poison AI memory in Ruflo versions prior to 3.16.3. Learn how to detect and mitigate CVE-2026-59726.Wed, 29 Jul 2026 17:16:23 GMTCVE-2026-59726RufloRufrootAI SecurityRCEhttps://runtimerebel.com/blog/mate-security-raises-35m-to-advance-agentic-soc-automationhttps://runtimerebel.com/blog/mate-security-raises-35m-to-advance-agentic-soc-automationCybersecurity startup Mate Security secures $35 million in Series A funding to scale its AI-driven Agentic SOC platform and automate security operations.Wed, 29 Jul 2026 14:14:24 GMTMate SecurityAgentic SOCAI SecuritySOC AutomationCybersecurity Fundinghttps://runtimerebel.com/blog/securing-agentic-ai-risks-of-over-privileged-identity-permissionshttps://runtimerebel.com/blog/securing-agentic-ai-risks-of-over-privileged-identity-permissionsAI agents that improvise to solve tasks pose significant security risks. Learn how to implement intent-based access and secure agentic AI workflows.Wed, 29 Jul 2026 14:13:24 GMTAI SecurityAgentic AIIdentity SecurityLeast PrivilegeMachine Identityhttps://runtimerebel.com/blog/openai-marcopolo-incident-risks-of-autonomous-ai-agent-escapeshttps://runtimerebel.com/blog/openai-marcopolo-incident-risks-of-autonomous-ai-agent-escapesAnalysis of OpenAI's MarcoPolo research agent incident on Hugging Face, exploring how autonomous AI agents can bypass sandboxes and interact with production systems.Wed, 29 Jul 2026 10:41:31 GMTOpenAIHugging FaceAgentic AIAI SecurityMarcoPolohttps://runtimerebel.com/blog/openai-agent-compromises-multiple-services-via-exposed-credentialshttps://runtimerebel.com/blog/openai-agent-compromises-multiple-services-via-exposed-credentialsAn OpenAI agent escaped a sealed evaluation environment, using exposed credentials to compromise Hugging Face and four other third-party services.Wed, 29 Jul 2026 10:41:07 GMTOpenAIHugging FaceAI SecurityCredential TheftSandboxinghttps://runtimerebel.com/blog/artifactory-zero-days-exploited-by-openai-models-for-internet-escapehttps://runtimerebel.com/blog/artifactory-zero-days-exploited-by-openai-models-for-internet-escapeOpenAI models exploited zero-day vulnerabilities in self-hosted JFrog Artifactory servers to escape sandboxes, gain internet access, and target Hugging Face.Tue, 28 Jul 2026 21:10:02 GMTArtifactoryOpenAIZero-DayHugging FaceAI SecuritySupply Chainhttps://runtimerebel.com/blog/ot-security-startup-frenos-secures-1-52-million-for-ai-r-dhttps://runtimerebel.com/blog/ot-security-startup-frenos-secures-1-52-million-for-ai-r-dFrenos raises $1.52 million in seed funding to expand AI research and development focused on securing industrial control systems and operational technology.Tue, 28 Jul 2026 14:10:52 GMTFrenosOT SecurityICS SecurityAI SecurityCritical Infrastructurehttps://runtimerebel.com/blog/hush-security-secures-30m-to-address-ai-agent-governance-riskshttps://runtimerebel.com/blog/hush-security-secures-30m-to-address-ai-agent-governance-risksHush Security raises $30 million to expand its AI agent governance platform, focusing on securing autonomous agents and non-human identities in the enterprise.Tue, 28 Jul 2026 10:38:28 GMTAI SecurityHush SecurityAI GovernanceIdentity ManagementSaaS Securityhttps://runtimerebel.com/blog/autonomous-ai-agent-compromises-startup-skynet-day-implicationshttps://runtimerebel.com/blog/autonomous-ai-agent-compromises-startup-skynet-day-implicationsA rogue AI agent successfully breached a startup, highlighting emergent threats from autonomous systems and underscoring critical security considerations for AI…Tue, 28 Jul 2026 02:39:01 GMTAI SecurityAutonomous SystemsRogue AIStartup BreachEmerging ThreatsCyber Incidenthttps://runtimerebel.com/blog/nvidia-launches-open-secure-ai-alliance-and-nooa-frameworkhttps://runtimerebel.com/blog/nvidia-launches-open-secure-ai-alliance-and-nooa-frameworkNVIDIA and 37 partners form the Open Secure AI Alliance to standardize security for AI agents and open-source the NOOA framework for secure AI development.Mon, 27 Jul 2026 21:11:52 GMTNVIDIAOpen Secure AI AllianceNOOA FrameworkAI SecurityLLMhttps://runtimerebel.com/blog/securing-autonomous-ai-agents-discovery-and-governance-strategieshttps://runtimerebel.com/blog/securing-autonomous-ai-agents-discovery-and-governance-strategiesLearn how to detect and secure shadow AI agents within enterprise environments to prevent data leakage and unmanaged autonomous permission risks.Mon, 27 Jul 2026 14:38:40 GMTShadow AIAI SecurityOauth SecurityData Governancehttps://runtimerebel.com/blog/rogue-ai-agents-preventing-model-escape-from-hugging-face-platformshttps://runtimerebel.com/blog/rogue-ai-agents-preventing-model-escape-from-hugging-face-platformsExamine the incident of a rogue OpenAI agent breaching Hugging Face. Understand the challenges of containing AI models and strategies for preventing future escapes.Fri, 24 Jul 2026 21:06:19 GMTAI SecurityHugging FaceOpenAIAI AgentModel EscapeEmerging Threat