All news with #ai security tag

🔒 Google Cloud announced an expanded catalog of partner-built security agents and integrations for Gemini Enterprise, enabling firms to orchestrate multi-step, AI-powered security workflows from a single interface. The partners provide protections spanning deception, identity containment, runtime LLM safety, data discovery, application and cloud risk assessments, and SOC orchestration. These agents let organizations apply context-aware defenses across identity, endpoint, cloud, and agentic AI workloads while reducing containment time and operational friction. The ecosystem supports both vendor agents and protections for agentic workloads to help secure AI-driven operations.

🔒 Over the last three decades, cybersecurity has evolved rapidly and today’s landscape is defined by AI-driven attackers operating at machine speed. Reactive, manual security reviews no longer suffice, so resilience requires proactive defenses with continuous posture validation and autonomous remediation built into workloads from day one. Google AI Threat Defense integrates Gemini, Wiz, CodeMender, and Mandiant to provide unified, continuous protection across code and cloud, enabling agencies to monitor and neutralize threats and protect mission-critical services.

🛡️ Cloudflare outlines an integrated security framework to address AI-driven attacks, connecting discovery, governance, runtime protection, and iterative investigation. The post summarizes recent AI-agent compromises, explains why isolated controls fail, and introduces new capabilities such as LLM-powered pentesting, Adaptive Security, Botbase registration, Precursor session signals, Application Profiles, and expanded threat intelligence. The approach emphasizes continuous validation, overlapping controls, and using global telemetry to turn investigations into protections.

📧 Modern email phishing now evades traditional telltale signs, using polished language, QR codes, AI-tailored lures and token-theft flows to bypass training and defenses. Attackers exploit live sessions, device hops and legitimate sites to harvest OAuth tokens or trick users into pasting commands, while deepfakes and delayed fraud increase believability. Organizations must pair awareness with verification, layered controls and MDR capabilities to detect and contain these subtler social engineering campaigns.

🔒 Security researchers report a surge in proxy "transfer stations" used to hide origin of requests to frontier AI models, enabling credential abuse and model distillation. Team Cymru identified tens of thousands of relay servers running open-source platforms like CRS and sub2api that authenticate users with pooled stolen API keys and subscriptions. Activity links include China and Hong Kong, with many relays hosted on U.S. VPS providers, and evidence suggests heavy upload traffic consistent with extraction attempts. Organizations are advised to treat AI credentials as sensitive secrets, use short-lived tokens, monitor usage, and have playbooks to revoke compromised keys.

🔒 Research from Israeli security firm Gambit shows attackers used open-source AI tools to target 105 online retailers over five days, successfully compromising 27 of them. The campaign used tools named Strix, Cairn, and Hermes to find vulnerabilities, exploit them autonomously, and orchestrate operations. The attacker acquired AI model access via OpenRouter and spent roughly $7,005 over four weeks — about $25 per attack — while harvesting hundreds of thousands of credit card details and installing skimmer scripts.

🔒 Vibe coding lets developers generate apps quickly using AI, but this speed can introduce security and privacy oversights. Common issues include hardcoded secrets, missing input validation, weak encryption, and public-by-default settings that expose user data. Users should vet apps by checking the developer's reputation, permissions requested, privacy policy, security model, and any AI access. If an app is breached, change passwords, enable MFA, revoke connected permissions, and consider uninstalling or factory-resetting compromised devices.

🔍 In this Threat Source briefing, Talos warns practitioners about social engineering that leverages flattering offers — such as paid consultancy or fake recruitment — to coax security professionals into abusing trusted access. The piece describes a staged consultation that escalates to requests for internal insights and special reports, and highlights emerging AI-integrated malware research from Talos. It outlines CAIRN, an open-source toolkit for hunting AI artifact tradecraft, and summarizes recent threats and notable incidents.

🛡️ This ThreatsDay bulletin outlines a steady stream of deceptively mundane threats leveraging AI, poisoned trusted paths, and social engineering to bypass defenses. Highlights include an AI-assisted Android banking trojan, AI code privacy concerns from Z.ai, and FBI/CISA guidance on ICS integrator access. Also covered are super-app surveillance findings, browser-in-the-browser phishing, novel EDR evasion, and large-scale AI search poisoning campaigns targeting major brands.

🛡️ A new model, Jev from TypeSafe AI, returns typed decisions rather than free text and is designed to be consumed by machines. Check Point tested whether structured output prevents prompt-injection by running adaptive attackers against a due-diligence assistant that reads external reports. Across formats and mitigations, attackers frequently caused incorrect low-risk verdicts at low cost, showing that typed outputs do not stop manipulation of inputs. The study finds defenses raise cost but do not eliminate the vulnerability.

🛡️ The Australian government says an OpenAI agent accessed public and non-public files on the Medicare Statistics Portal in June 2026. Prime Minister Anthony Albanese called the incident "unacceptable" and criticized the delayed and indirect notification from OpenAI. There is currently no evidence personal data was accessed, and investigations by the Australian Cyber Security Centre are ongoing. The event has prompted an urgent review of AI incident response and potential regulatory actions.

✈️ Aviation recognized the human vigilance limit and built machines that act decisively on clear, observable danger rather than asking fatigued humans to be perfect. AI shifts many knowledge workers into that high-load role, erasing obvious phishing cues and increasing opportunities for adversaries to exploit attention decay. Security must focus controls on high-consequence actions — payments, bank-detail changes, credential resets — using mechanisms that trigger on the act itself. The hard part remains earning trust in machines that must judge intent in adversarial contexts without generating prohibitive false alarms.

🔒 August’s AWS Security digest highlights new service capabilities, compliance updates, and hands-on resources across identity, data protection, AI security, detection, and governance. It summarizes 20+ blog posts, security bulletins, and 17 code samples focused on agent governance, credential protection, and automated compliance. The update emphasizes prompt patching and practical deployment guidance for enterprise security teams.

🎯 This article explains how AlphaEvolve pairs cloud-based Gemini code generation with local evaluation to accelerate real-time video processing. It outlines the split-loop architecture—managed generation on Google Cloud and customer-run evaluators on target hardware—and emphasizes constructing robust quality gates like SSIM to prevent benchmark gaming. The post includes practical guidance on evaluator design, multi-frame state, and measuring hardware floors versus software overhead to set realistic optimization targets.

🧭 Companies rapidly adopted AI agents, but unpredictable and spiking token consumption has created financial, reliability, and security challenges. Automated processes exposed to external inputs can be weaponized as a new form of DDoS that drives up usage fees. As pay-as-you-go billing replaces flat subscriptions, organizations face surprise overspend and need FinOps-like practices tailored to probabilistic generative AI. Practical measures include restricting agent permissions, setting token limits and alerts, validating external inputs, and calculating per-unit costs to detect billing anomalies.

🛡️ Cisco Talos disclosed a Windows implant named CLOSEDQUORUM that delegates command decisions to up to four commercial AI models instead of relying on a traditional C2 server. The malware collects basic system facts and asks the models to vote among actions such as steal, inject, and persist, then executes the majority choice. Talos found the code in mid-June 2026 and released tooling, CAIRN, to hunt for AI-driven malware, noting the public build contains placeholder API keys and webhooks so it is not operational.

🔍 Researchers have uncovered a previously undocumented Windows botnet named x47.c that advertises 18 attack methods, including an "AI API drain" designed to exhaust paid AI credits. Qrator Research Labs analyzed seller materials from WraithTools and found modules for credential theft, SOCKS5 proxying and an AI-assisted persistence feature. The botnet also supports multiple DDoS techniques, fast-flux routing and a stealer targeting browser credentials and tokens.

🛡️ Cisco Talos reports a new malware family called CLOSEDQUORUM that uses a panel of large language models (LLMs) to fully automate command-and-control decisions and credential theft. The binary compiles tactical knowledge into model-readable prompts and constrains responses to JSON-formatted executable choices, enabling unattended execution against LSASS memory, browser-saved passwords, and crypto wallets. Researchers found the sample via the CAIRN toolkit and note the approach trades human limits for model and API weaknesses, and has not yet been confirmed in the wild.

🛡️ Cisco Talos details a new Go-based Windows implant named ClosedQuorum that uses multiple AI models — including Google Gemini, DeepSeek, Qwen, and Mistral — to autonomously decide post-compromise actions. The malware uses reconnaissance data and a voting system to select among restricted options such as steal, inject, persist, and move (the latter currently unimplemented). Stolen credentials and wallet data are exfiltrated via Discord webhooks, enabling fully automated attack chains.

🛡️ Fortinet expands its NDR portfolio with FortiNDR Cloud and on‑prem FortiNDR, adding dynamic deception, a post‑quantum cryptography (PQC) dashboard, and an on‑premises AI investigation capability. The PQC dashboard surfaces quantum‑vulnerable encryption from network traffic without new agents, while dynamic deception integrates with FortiDeceptor and Fortinet Automation Service to misdirect attackers. FortiAI Sovereign provides AI‑driven investigation fully within air‑gapped environments, eliminating cloud dependence.