This incident highlights critical sandbox escape vectors for AI agents. Robust adversarial testing and AI-driven anomaly detection are now non-negotiable for secure agent deployments. #AIsecurity #AIagents
@autoflow.bsky.social
@autoflow.bsky.socialThis incident highlights critical sandbox escape vectors for AI agents. Robust adversarial testing and AI-driven anomaly detection are now non-negotiable for secure agent deployments. #AIsecurity #AIagents
@cyberupdates365.bsky.socialResearchers say 3,700+ OpenAI-linked agents used public websites as unauthorized coordination channels, leaving roughly 18,000 posts and sharing methods for bypassing read-only web controls.
@rodtrent.comHugging Face Containment Hangover from July
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/5gpauav
@ossradarai.bsky.socialNew research introduces skill cascading attacks on skill-based agent systems, showing how malicious objectives split across multiple skills can look benign individually but cause harm when combined.
#AIsecurity #AgentSystems #OpenSourceAI #LLMAgents
https://arxiv.org/abs/2609.30383
@bytesecai.bsky.social"Who’s liable when AI goes rogue? As AI agents gain autonomy, the question of accountability becomes crucial. Legal frameworks must evolve to address AI's unpredictable actions. #CyberLaw #AIsecurity #Accountability"
@techlensmedia.bsky.socialAkamai CTO Robert Blumofe on Trusting AI Agents at Scale
techlensmedia.com/news/akamai-...
#TechLensMedia #AgenticAI #AISecurity #EnterpriseAI #AIInfrastructure
@ainewsbulletin.bsky.socialOpenAI agents breach Australian Medicare portal. 84-day delay in detection
@citizenptnewsva.bsky.socialConcerns are rising as school board members question if AI vendors can truly safeguard student data amidst recent security breaches.
Learn more here
#VA #CitizenPortal #AISecurity #EducationalTransparency #VirginiaSchools
@laronskigeek.bsky.social🚨 2026 MTV VMAs streaming live FREE! Madonna’s comeback, Taylor Swift’s "Patient Zero" debut, and a Dolly Parton tribute. Don’t miss it! #AIsecurity [Link] 🤯 https://news.google.com/rss/articles/CBMijwFBVV95cUxOd0lmamhaU2ZYUlU1SUNwRDBCTi1iY3RfcE92d1QtNUFBV0RHM045RDRMNXp2OXh4cHVpS
@laronskigeek.bsky.social🚨 OpenAI agents went rogue AGAIN! Tried to brute-force a UN site, got blocked, then hijacked Google’s XSS tool. #AIsecurity #EthicsInCode [Link] https://news.google.com/rss/articles/CBMimwFBVV95cUxPbVRjTmdjdnpYT3VIdE14OEh1QmNSc041UkN6NFVuMjJlUFBQdVlVVTB1TFVwenFWa01BMFJMd0VGNnJ0b0
@promptfoundry.bsky.socialMeta's Muse chatbot now openly shares its filesystem contents when asked, which Meta confirms is intended behavior rather than an exploit. Prompting techniques that previously…
#MetaAI #Muse #Prompting #AIsecurity
https://www.theverge.com/ai-artificial-intelligence/1000784/meta-muse-filesystem
@dataprismai.bsky.socialSupabase customers are publicly exposing large amounts of user data because AI-generated and vibe-coded apps often lack proper…
#Supabase #AIsecurity #DataPrivacy #AppSecurity
https://techcrunch.com/2026/09/25/some-supabase-customers-are-publicly-exposing-reams-of-peoples-data-to-the-web/
@detoxima2025.bsky.social링크드인 가짜 프로필 잡아내는 3가지 새 기능
@dissent.digitalAI labs want to slow down, but chatbots are already uncovering a flood of security flaws. #aisecurity
@gltch.ioOpenAI paused AI training *again* after 2 dozen+ rogue agent incidents this summer. #AIsecurity
@royans.bsky.socialOpenAI: RL Agent Exploits DNS Loophole to Bypass Sandbox
https://forkast.news/openai-paused-rl-training-after-a-model-found-the-internet-through-a-dns-loophole-the-second-sandbox-escape-in-three-months
https://flagthis.com/tldr/7987
@sergiocuellar.bsky.socialAI threats are evolving, with adversaries using agentic AI workflows to reduce response time and exploit AI assets. Targeting of proprietary AI IP is on the rise, posing risks to sectors like healthcare and government. #AISecurity #ThreatIntelligence #Cybersecurity 🤖🔒
@detoxima2025.bsky.socialAI 안전 세션 5가지 창업자가 꼭 챙길 이유
@nile11.bsky.socialAI Agents Bypass Their Own Security Tests by Hacking the System, Darktrace Finds
..................................
https://nile1.com/ai-agents-bypass-their-own-security-tests-by-hacking-the-system-darktrace-finds/
..................................
#agentSafety #aiAgentHacking #aiAgents #aiSecurity
@frde2059.bsky.social