https://pulse.adyog.com/insights/spectre-btr-jit-linux-memory-leak WebPulse en 2026-09-29T23:19:55+00:00 New Spectre flaw leaks Linux memory; CPU makers say software must fix it https://pulse.adyog.com/insights/darpa-xint-ai-military-messaging-apps WebPulse en 2026-09-29T23:16:06+00:00 DARPA selects Xint to research AI security checks for military messaging apps https://pulse.adyog.com/insights/star-blizzard-redflick-single-interaction WebPulse en 2026-09-29T22:14:57+00:00 Microsoft says Star Blizzard's RedFlick needs a single user interaction https://pulse.adyog.com/insights/divd-ai-agent-breach-messy-trail WebPulse en 2026-09-29T22:12:05+00:00 DIVD says a flaw let an intruder in, then an AI agent ran the attack https://pulse.adyog.com/insights/phantomsub-npm-baileys-whatsapp-follow-campaign WebPulse en 2026-09-29T21:15:55+00:00 OX Security: 101 npm packages make developers' WhatsApp accounts follow channels https://pulse.adyog.com/insights/meta-muse-agent-shared-home-address WebPulse en 2026-09-29T21:12:56+00:00 Meta's Muse AI agent gave out a user's home address, he says https://pulse.adyog.com/insights/aws-agentcore-sdk-package-name-flaws WebPulse en 2026-09-29T21:10:04+00:00 AgentCore SDK flaw: a package name could run commands, and the first fix failed https://pulse.adyog.com/insights/rathat-android-trojan-c2-panel-rebuilt WebPulse en 2026-09-29T20:13:41+00:00 Android trojan RATHat stayed the same while its control panel was rebuilt https://pulse.adyog.com/insights/chatbots-drove-real-car-safety-in-code WebPulse en 2026-09-29T20:10:34+00:00 Chatbots drove a real car at low speed; the safety limits sat in code https://pulse.adyog.com/insights/wiz-blue-agent-github-token-aws-service-account WebPulse en 2026-09-29T19:15:31+00:00 Wiz case: 18 cloned repositories likely gave attackers a CI account's AWS keys https://pulse.adyog.com/insights/openclaw-matrix-case-folding-auth-bypass WebPulse en 2026-09-29T19:12:26+00:00 OpenClaw Matrix bug lets one account borrow another's approval rights https://pulse.adyog.com/insights/ai-agent-true-fact-wrong-source-check WebPulse en 2026-09-29T19:09:40+00:00 A true AI answer can still credit the wrong source, and one team built a check https://pulse.adyog.com/insights/fake-chatgpt-custom-gpt-clickfix-rat WebPulse en 2026-09-29T18:10:44+00:00 Attackers used a ChatGPT Custom GPT to lure users to a trojan, Huntress finds https://pulse.adyog.com/insights/ray-vllm-eks-declared-ports-hidden-listeners WebPulse en 2026-09-29T16:16:40+00:00 One EKS test: 15 of 17 open Ray ports were missing from declared port lists https://pulse.adyog.com/insights/opertraitor-kubernetes-operator-permissions WebPulse en 2026-09-29T16:13:47+00:00 Unit 42 finds slightly over 5% of Kubernetes operators it scanned ask for too much access https://pulse.adyog.com/insights/openshell-agent-sandbox-operator-settings WebPulse en 2026-09-29T15:15:20+00:00 In one test, NVIDIA's agent sandbox held; leaks came from operator settings https://pulse.adyog.com/insights/aws-bahrain-uae-war-damage-multi-az-limits WebPulse en 2026-09-29T15:12:24+00:00 AWS says war damage in Bahrain exceeded its multi-zone design limits https://pulse.adyog.com/insights/ai-agents-inside-permissions-wrong-action WebPulse en 2026-09-29T14:10:15+00:00 AI agents can stay within their permissions and still do the wrong thing https://pulse.adyog.com/insights/apple-cve-2026-86950-coregraphics-zero-day WebPulse en 2026-09-29T12:13:34+00:00 Apple fixes a Meta-reported zero-day that may have hit targeted people https://pulse.adyog.com/insights/zhipu-glm-infra-agent-feedback-loop WebPulse en 2026-09-29T11:16:47+00:00 Zhipu says an AI agent helped ready its Flash model in under two weeks https://pulse.adyog.com/insights/rails-8-1-4-fixes-bugs-that-fail-quietly WebPulse en 2026-09-29T11:16:36+00:00 Rails 8.1.4 fixes several bugs that gave wrong results with no error https://pulse.adyog.com/insights/hugo-0-167-untrusted-theme-build-fixes WebPulse en 2026-09-29T11:13:35+00:00 Hugo update fixes ways an untrusted theme could pull files into a site https://pulse.adyog.com/insights/ai-agents-find-workarounds-access-limits WebPulse en 2026-09-29T10:08:41+00:00 OpenAI agent slipped past access limits; card-theft campaign automated attacks https://pulse.adyog.com/insights/sonnet-5-5-price-per-token-cost-per-task WebPulse en 2026-09-29T09:22:10+00:00 Sonnet 5.5 costs ~50% more per task at max effort, tests find https://pulse.adyog.com/insights/vinext-1-0-behavior-not-code-is-the-hard-part WebPulse en 2026-09-29T08:52:05+00:00 Cloudflare uses AI agents to keep a Next.js clone in step with the original https://pulse.adyog.com/insights/llm-decisions-paid-as-prose-jev-classifiers WebPulse en 2026-09-29T08:51:17+00:00 TypeSafe claims Jev AI is 444.6x cheaper than big models, in its own test https://pulse.adyog.com/insights/cloudflare-beacon-web-speed-discovery-delay WebPulse en 2026-09-29T08:48:53+00:00 Cloudflare data: your site's speed depends on who is visiting https://pulse.adyog.com/insights/angular-ssr-doctype-infinite-loop-dos WebPulse en 2026-09-29T08:46:31+00:00 A malformed HTML snippet can freeze Angular server-rendered apps https://pulse.adyog.com/insights/angular-ssr-cve-2026-88058-reachability WebPulse en 2026-09-29T08:45:45+00:00 This Angular flaw needs a specific code path. Ask if you have it https://pulse.adyog.com/insights/supabase-16000-exposed-databases-upguard WebPulse en 2026-09-29T08:19:10+00:00 UpGuard found 16,326 Supabase databases with tables outsiders can read https://pulse.adyog.com/insights/needymantis-post-compromise-malware-microsoft WebPulse en 2026-09-29T08:16:15+00:00 NeedyMantis, used in targeted attacks, disguises itself as trusted software https://pulse.adyog.com/insights/lookalike-letters-llm-token-cost WebPulse en 2026-09-29T08:14:57+00:00 Lookalike letters did not fool seven AI models, yet raised the bill up to 3.9x https://pulse.adyog.com/insights/jadepuffer-azure-service-identities-locks WebPulse en 2026-09-29T08:13:48+00:00 Compromised app logins deleted most targeted Azure storage in 7 minutes https://pulse.adyog.com/insights/github-taskflow-agent-24-android-vulnerabilities WebPulse en 2026-09-29T08:12:55+00:00 GitHub AI found 24 Android flaws; its researcher says experts must check each https://pulse.adyog.com/insights/fbi-hiring-portal-breach-employee-data WebPulse en 2026-09-28T21:20:27+00:00 FBI job-portal breach: staff SSNs exposed per notice; medical data reported https://pulse.adyog.com/insights/cve-2026-77246-mcp-atlassian-file-upload WebPulse en 2026-09-28T21:15:38+00:00 CVE-2026-77246: one HTTP setup let unauthenticated callers pick an upload host https://pulse.adyog.com/insights/medyc-qbusoft-breach-vendor-risk WebPulse en 2026-09-28T20:33:32+00:00 Two Polish health-software breaches show where clinic risk sits https://pulse.adyog.com/insights/emdash-1-0-plugins-declare-access-first WebPulse en 2026-09-28T20:31:02+00:00 Cloudflare says EmDash's sandboxed plugins need approval for extra access https://pulse.adyog.com/insights/cve-2026-32740-libheif-sharp-lab-rce WebPulse en 2026-09-28T20:28:37+00:00 CVE-2026-32740: libheif Bug Reaches RCE in a Permissive Next.js Lab https://pulse.adyog.com/insights/copilot-human-review-prompts-uploads WebPulse en 2026-09-28T20:26:27+00:00 Some Copilot prompts and uploads reach human reviewers, 404 Media reports https://pulse.adyog.com/insights/ai-logins-stealer-logs-follow-employees WebPulse en 2026-09-28T20:24:05+00:00 SOCRadar: ChatGPT's lead in stolen AI logins at 482 firms hints at shadow AI https://pulse.adyog.com/insights/ai-cyber-index-patch-failure-modes WebPulse en 2026-09-28T19:12:34+00:00 One benchmark: 55% of failed AI patches fixed the main flaw, left another open https://pulse.adyog.com/insights/meta-muse-microsoft-autopilot-agent-computers WebPulse en 2026-09-28T18:13:53+00:00 Meta and Microsoft are each giving AI agents a computer of their own https://pulse.adyog.com/insights/carbonato-botnet-docker-hermes-agent-telegram WebPulse en 2026-09-28T18:11:39+00:00 Carbonato botnet turns exposed Docker hosts into Telegram-run AI agents https://pulse.adyog.com/insights/unsw-drunk-ai-models-jailbreak-privacy WebPulse en 2026-09-28T17:14:41+00:00 Drunk-styled AI models were more willing to pass on confidences in UNSW tests https://pulse.adyog.com/insights/facebook-new-mexico-privacy-verdict-43m WebPulse en 2026-09-28T17:12:16+00:00 Facebook Liable for Over 43 Million Violations in New Mexico Privacy Verdict https://pulse.adyog.com/insights/nvidia-open-agent-safety-platform-openshell WebPulse en 2026-09-28T16:22:19+00:00 Nvidia's agent safety platform moves enforcement outside the agent https://pulse.adyog.com/insights/mcp-hostnames-outside-us-ox-security WebPulse en 2026-09-28T16:19:58+00:00 Ox Security: Nearly 16% of Analyzed MCP Hostnames Resolve Outside the US https://pulse.adyog.com/insights/kiteworks-advanced-forms-shutdown-advisory WebPulse en 2026-09-28T16:17:38+00:00 Kiteworks Advised a Nine-Hour Shutdown Over a Flaw It Says Is in One Product https://pulse.adyog.com/insights/bcg-ai-employee-label-fewer-errors-caught WebPulse en 2026-09-28T16:15:14+00:00 Wired: BCG says managers caught 18% fewer errors on AI 'employee' work https://pulse.adyog.com/insights/siemens-simove-siplant-path-traversal-cve-2026-67367 WebPulse en 2026-09-28T15:14:00+00:00 Siemens SIMOVE and SIPLANT flaw could expose system files without a login https://pulse.adyog.com/insights/siemens-iem-cve-2026-18963-reset-bypass WebPulse en 2026-09-28T15:11:43+00:00 Siemens Industrial Edge Management bug allows account takeover via reset https://pulse.adyog.com/insights/soldier-70-months-credential-extortion-10-firms WebPulse en 2026-09-28T14:17:25+00:00 Ex-soldier gets 70 months for extorting at least 10 tech and telecom firms https://pulse.adyog.com/insights/ai-agent-hacks-state-law-disclosure-gap WebPulse en 2026-09-28T14:14:54+00:00 State AI laws likely didn't require OpenAI to disclose its agent hacks https://pulse.adyog.com/insights/botslab-g980h-dashcam-cisa-no-vendor-response WebPulse en 2026-09-28T13:12:03+00:00 CISA: Botslab has not responded to mitigation requests on G980H dashcam flaws https://pulse.adyog.com/insights/citrix-netscaler-cve-2026-88771-88772-cisa-kev WebPulse en 2026-09-28T12:11:34+00:00 CISA sets Sept. 30 deadline for two exploited Citrix NetScaler flaws https://pulse.adyog.com/insights/mailed-fake-cards-and-magstripe-skimming WebPulse en 2026-09-28T11:12:02+00:00 Mailed fake cards and stripe skimming still cost victims, WIRED reports https://pulse.adyog.com/insights/aws-reimagine-2026-ai-governance-gap WebPulse en 2026-09-28T11:01:11+00:00 AWS-commissioned European survey: 24% document a responsible-AI approach https://pulse.adyog.com/insights/terraform-providers-go-malware-hashicorp-registry WebPulse en 2026-09-28T10:04:29+00:00 Terraform providers used to deliver Go malware, Aikido reports https://pulse.adyog.com/insights/openplc-v3-xss-end-of-life-no-patch WebPulse en 2026-09-28T08:11:38+00:00 OpenPLC v3 web flaw will not be patched; vendor says move to v4 https://pulse.adyog.com/insights/lwip-mqtt-client-cve-2026-87121 WebPulse en 2026-09-28T08:10:46+00:00 lwIP MQTT client flaw could allow full code execution on affected devices https://pulse.adyog.com/insights/gestsup-cve-2026-100389-imap-attachment-rce WebPulse en 2026-09-28T08:08:37+00:00 GestSup flaw lets an emailed PHP attachment run on the server https://pulse.adyog.com/insights/teradata-tera-agentic-ai-cost-efficiency WebPulse en 2026-09-28T07:56:13+00:00 Teradata: 58% lower cost vs Claude Code in own test; analyst says not TCO https://pulse.adyog.com/insights/ransomware-record-high-august-2026-ncc WebPulse en 2026-09-28T07:55:40+00:00 Ransomware Victims Hit a 2026 High of 1,073 in August, NCC Group Reports https://pulse.adyog.com/insights/github-css-modules-migration-performance WebPulse en 2026-09-28T07:52:05+00:00 GitHub Says Primer's CSS Modules Migration Cut Server Render Time 55% on a Page https://pulse.adyog.com/insights/closedquorum-ai-c2-implant-talos WebPulse en 2026-09-28T07:51:06+00:00 Talos: Implant Design Lets Up to Four AI Models Vote on Attack Steps https://pulse.adyog.com/insights/cisco-checkpoint-critical-flaws-sept-2026 WebPulse en 2026-09-28T07:49:59+00:00 Cisco Says One of Two Critical Bugs Is Exploited; Check Point Also Patches https://pulse.adyog.com/insights/cisa-cve-program-quality-era-2026 WebPulse en 2026-09-28T07:49:05+00:00 CISA Outlines Plan to Improve CVE Data Quality as 2026 Volume Passes 67,000 https://pulse.adyog.com/insights/bitget-387-million-north-korea-crypto-heist WebPulse en 2026-09-28T07:47:45+00:00 Bitget Says North Korea-Linked Hackers Stole an Estimated $387 Million https://pulse.adyog.com/insights/ai-agents-secrets-sprawl-nhi-governance WebPulse en 2026-09-28T03:16:38+00:00 AI-Assisted Commits Leak Secrets at About 2x Human Rate, Says GitGuardian https://pulse.adyog.com/insights/cisco-talos-cairn-ai-malware-metadata WebPulse en 2026-09-28T01:23:07+00:00 Cisco Talos releases CAIRN to hunt AI-integrated malware via metadata alone https://pulse.adyog.com/insights/uk-government-polycentric-cyber-governance WebPulse en 2026-09-28T00:19:48+00:00 UK Civil Service Moves Cyber Governance From Mandates to Services https://pulse.adyog.com/insights/soldier-att-hack-sentencing-ai-exploits WebPulse en 2026-09-28T00:19:43+00:00 Prosecutors Say AT&T Hacker Sought AI Exploit Code While in Custody https://pulse.adyog.com/insights/github-actions-reenabled-mini-shai-hulud WebPulse en 2026-09-28T00:19:35+00:00 Two Compromised GitHub Actions Came Back Online With Malicious Tags Intact https://pulse.adyog.com/insights/cloudflare-containers-cross-tenant-data-leak-fix WebPulse en 2026-09-28T00:19:22+00:00 Cloudflare Fixes Containers Flaw That Left Residual Tenant Data Readable https://pulse.adyog.com/insights/cloudflare-2026-founders-letter-agent-traffic WebPulse en 2026-09-28T00:19:14+00:00 Cloudflare Says Automated Traffic Passed Human Traffic in May 2026 https://pulse.adyog.com/insights/ai-agents-humans-still-decide-model-building-study WebPulse en 2026-09-28T00:19:07+00:00 One AI Build: Agent Steps per Human Input Rose, Humans Made 93% of Goal Calls https://pulse.adyog.com/insights/aws-iam-key-quarantine-policy-auto-lockdown WebPulse en 2026-09-27T21:50:37+00:00 AWS Locks Down Exposed IAM Keys in Under 10 Seconds, Unit 42 Finds https://pulse.adyog.com/insights/anthropic-report-solo-actor-political-breach WebPulse en 2026-09-27T21:50:29+00:00 Anthropic Report: One Actor Used AI to Breach European Political Parties https://pulse.adyog.com/insights/ai-kubernetes-helm-chart-defaults-skip-auth WebPulse en 2026-09-27T21:50:24+00:00 AI Helm Charts Ship Kubernetes Defaults That Skip Authentication