We don't need to hack your AI Agent to hack your AI Agent …and we don't need an AI agent for that either :)

Via a large enterprise's AI assistant, we obtained access to several million Entra identities and all chat logs including attachments — no prompt injection or model tricks required.

For all we know, the poor agent was not at fault and may not have even been able to witness what was happening.

https://srlabs.de/blog/hacking-ai-agent

#AI #AIhacking #VulnerabilityDisclosure #ResponsibleDisclosure

The Ivanti EPMM zero-days underline a recurring issue: edge-facing management platforms remain prime targets.

Confirmed incidents across EU institutions and government bodies show how quickly critical vulnerabilities can be weaponized. Even without confirmed device compromise, exposure of management infrastructure carries systemic risk.

Source: https://www.technadu.com/ivanti-zero-day-vulnerabilities-exploited-in-global-cyberattacks-dutch-government-breached-possibly-european-commission/619761/

💬 Are edge device vulnerabilities becoming the dominant zero-day threat class?

🔔 Follow @technadu for ongoing vulnerability and incident analysis

#InfoSec #ZeroDay #Ivanti #MDM #EdgeSecurity #VulnerabilityDisclosure #CyberDefense #TechNadu

Acknowledging Reality in Vulnerability Disclosure.

Every few years, vulnerability disclosure is declared settled. We are told that the ecosystem has matured, that coordinated disclosure is the answer, and that whatever remains outside this model is either irresponsible, obsolete, or simply irrelevant.

🔗 https://www.foo.be/2026/02/Acknowledging-Reality-in-Vulnerability-Disclosure

#vulnerabilitymanagement #gcve #cve #cybersecurity #cvd #vulnerabilitydisclosure #open

This is my bloody personal blog, not an official statement. .

PoC exploits are now public for CVE-2025-69258 in Trend Micro Apex Central (on-premise), a vulnerability that could allow unauthenticated RCE on affected systems.

A patch is available, and there are no confirmed exploitation reports so far. Public PoCs, however, tend to accelerate attacker interest.

Follow @technadu for objective and technically grounded infosec updates.

Source: https://www.helpnetsecurity.com/2026/01/08/trend-micro-apex-central-cve-2025-69258-rce-poc/

#Infosec #VulnerabilityDisclosure #PatchManagement #RCE #EnterpriseSecurity #ThreatLandscape

n8n has disclosed a CVSS 10.0 vulnerability that could lead to authenticated remote code execution under certain conditions.

The issue affected both self-hosted and cloud deployments and has been addressed in a patched release. Temporary mitigations include reducing untrusted access and limiting high-risk nodes.

Another reminder that workflow automation platforms require the same threat modeling as core infrastructure.

Source: https://thehackernews.com/2026/01/n8n-warns-of-cvss-100-rce-vulnerability.html

Follow @technadu for objective infosec updates.

#Infosec #RCE #VulnerabilityDisclosure #OpenSourceSecurity #CloudRisk #DevSecOps

Check out ˗ˏˋ ⭒ https://lnkd.in/gE2wUqgc ⭒ ˎˊ˗ to see my intro whilst you listen.

I'm thus re-naming this work as "CVE Keeper - Security at x+1; rethinking vulnerability management beyond CVSS & scanners". I must also thank @andrewpollock for reviewing several of my verbose drafts. 🫡

So, Security at x+1; rethinking vulnerability management beyond CVSS & scanners -

Most vulnerability tooling today is optimized for disclosure and alert volume, not for making correct decisions on real systems. CVEs arrive faster than teams can evaluate them, scores are generic, context arrives late, and we still struggle to answer the only question that matters: does this actually put my system at risk right now?

Over the last few years working close to CVE lifecycle automation, I’ve been designing an open architecture that treats vulnerability management as a continuous, system-specific reasoning problem rather than a static scoring task. The goal is to assess impact on the same day for 0-days using minimal upstream data, refine accuracy over time as context improves, reason across dependencies and compound vulnerabilities, and couple automation with explicit human verification instead of replacing it.

This work explores:

⤇ 1• Same-day triage of newly disclosed and 0-day vulnerabilities
⤇ 2• Dependency-aware and compound vulnerability impact assessment
⤇ 3• Correlating classical CVSS with AI-specific threat vectors
⤇ 4• Reducing operational noise, unnecessary reboots, and security burnout
⤇ 5• Making high-quality vulnerability intelligence accessible beyond enterprise teams

The core belief is simple: most security failures come from misjudged impact, not missed vulnerabilities. Accuracy, context, and accountability matter more than volume.

I’m sharing this to invite feedback from folks working in CVE, OSV, vulnerability disclosure, AI security, infra, and systems research. Disagreement and critique are welcome. This problem affects everyone, and I don’t think incremental tooling alone will solve it.

P.S.

• Super appreciate everyone that's spent time reviewing my drafts and reading all my essays lol. I owe you 🫶🏻
• ... and GoogleLM. These slides would have taken me forever to make otherwise.

Take my CVE-data User Survey to allow me to tailor your needs into my design - lnkd.in/gcyvnZeE
See more at - lnkd.in/gGWQfBW5
lnkd.in/gE2wUqgc

#VulnerabilityManagement #Risk #ThreatModeling #CVE #CyberSecurity #Infosec #VulnerabilityManagement #ThreatIntelligence #ApplicationSecurity #SecurityOperations #ZeroDay #RiskManagement #DevSecOps #CVE #CVEAnalysis #VulnerabilityDisclosure #SecurityData #CVSS #VulnerabilityAssessment #PatchManagement #AI #AIML #AISecurity #MachineLearning #AIThreats #AIinSecurity #SecureAI #OSS #Rust #ZeroTrust #Security

https://www.linkedin.com/feed/update/urn:li:activity:7409399623087370240

It's been a bit quiet over the last 24 hours, so it'll be a short post today, but we do have some positive news from Portugal regarding legal protections for security researchers. Let's dive in:

Portugal's New Cybercrime Law Protects Security Researchers 🛡️

- Portugal has updated its cybercrime law, introducing a legal safe harbour for security researchers who identify and responsibly disclose vulnerabilities.
- This exemption applies under strict conditions, including the sole purpose of improving cybersecurity, immediate reporting to affected parties and the CNCS, and avoiding disruptive or harmful techniques like DoS, social engineering, or data alteration.
- The move aligns Portugal with other nations like Germany and the US, which have also introduced similar protections to encourage ethical hacking and proactive vulnerability discovery without fear of legal repercussions.

🤖 Bleeping Computer | https://www.bleepingcomputer.com/news/security/portugal-updates-cybercrime-law-to-exempt-security-researchers/

#CyberSecurity #InfoSec #BugBounty #EthicalHacking #VulnerabilityDisclosure #CyberLaw #Portugal #RegulatoryUpdate

A 21-year-old cybersecurity entrepreneur in Russia has been arrested on treason charges, reportedly after publicly criticizing the Max messaging platform and raising concerns about new anti-cybercrime legislation.

With the case classified, details remain unclear - but the situation underscores the challenges faced by researchers operating in tightly regulated environments.

💬 What protections should security researchers have?
Follow @technadu for continuous global InfoSec coverage.

#InfoSec #Cybersecurity #DigitalRights #SecurityResearch #VulnerabilityDisclosure #TechNews

Testing out the new author attribution feature: https://skyplabs.com/posts/xss-iban-secured-transfer/

#Security #Web #WebSecurity #XSS #VulnerabilityDisclosure #Notary #Mastodon #Fediverse