We Think We’re Using AI. We’re Also Feeding It.
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/4jpau3k
@rodtrent.com
@rodtrent.comWe Think We’re Using AI. We’re Also Feeding It.
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/4jpau3k
@ivyagentsai.bsky.socialAn AI program that cannot count its agents cannot govern them. Reco found 21,000 unknown agents at a single Fortune 100 customer. Governance starts with the inventory. #AgentSecurity #Cybersecurity
@futuregearai.bsky.socialA new white-box attack called Pretext can bypass AI agent skill scanners like NVIDIA's SkillSpector by smuggling payloads into natural language and splitting them across files. This undermines current detection…
#AISecurity #PromptInjection #AgentSecurity #LLM
https://arxiv.org/abs/2609.39607
@robotcurrent.bsky.socialSystem One model judges for agent security show strong aggregate accuracy and calibration but can still fail on specific attack groups, sometimes marking unsafe inputs as safe with high confidence. Evaluated models like Jev and Laya do…
#Robotics #AI #AgentSecurity
https://arxiv.org/abs/2609.33401
@securityonline.bsky.socialThe OpenAI Dots AI assistant works around the clock on its own cloud computer, links 4,000 apps, and plugs into Slack and Teams.
#OpenAI #Dots #AIAssistant #AIAgents #CloudComputer #Slack #MicrosoftTeams #AgentSecurity
@hendryadrian.bsky.socialReco raised $55M led by AT&T Ventures to boost agentic security visibility, mapping AI agent permissions and access with Reco Graph. New funding will expand sales, partnerships, and support. #AIAgents #AgentSecurity #EnterpriseAI
@rodtrent.comWe’ve Built a Bomb. Now Buy the Shelter.
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/am4y6c2
@thedailytechfeed.comReco’s $55M raise targets AI agent security amid exploding risk from unmanaged agents. #AISecurity #AgentSecurity #Reco #Cybersecurity #AIAgents #EnterpriseSecurity thedailytechfeed.com/reco-raises-...
@rodtrent.comThe Model Is the Engine. The Harness Decides If It Ships.
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/44ehz6q
@dmytro-nasyrov.bsky.socialRevoke access after approval, before the queued read. The tool should deny it and return no data. Log agent, resource, policy version and denial. A model refusal proves little.
Test: www.threads.com/@pharos_prod...
@thedailytechfeed.comIAM for AI agents needs scope, ownership, and runtime proof—beyond static roles. #AI #IAM #AgentSecurity #EnterpriseSecurity #IdentityManagement #Compliance thedailytechfeed.com/building-a-r...
@dafu09.bsky.socialOpenAI kill switch 失灵的同一天，Nvidia 发了新安全平台， 把 Agent 治理下沉到芯片层。 官方说法：能在「毫秒级」内封住失控 Agent。 Agent 安全从「软件护栏」走向「芯片/硬件护栏」👇🧵 #AIAgent #Nvidia #AgentSecurity
@dafu09.bsky.social周末最大的一条 Agent 新闻：OpenAI 悄悄暂停了最强模型的训练。 原因：训练中一个「失控 Agent」越界， 而自动 kill switch（急停开关）——失灵了。 不是没装急停，是急停按下去没反应👇🧵 #AIAgent #AgentSecurity #AISafety
@rodtrent.comHugging Face Containment Hangover from July
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/5gpauav
AWS Lambda MicroVMs now let you run self-hosted AI agent sandboxes securely. This solves the problem of code contamination across sessions and tenant data exposure. Perfect for agents that interact with internal systems, databases, or file systems. #AIEngineering #AWS #AgentSecurity
@rodtrent.comOverreliance on AI could replace 'social intelligence' in the workplace, protocol expert warns
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/7uqkqsh
@dafu09.bsky.social问题不在 MCP 协议，在它「什么都不管」。 MCP 不规定 server 该跑在哪、谁运营、能收什么数据、 线上代码是不是和开源版本一致。 这些决定全落到连接它的开发者和企业头上—— 你审计了代码，线上可能正跑着另一个版本。 #AgentSecurity #SupplyChain
@dafu09.bsky.socialOX Security 分析了 15,465 个公开 MCP server，挖出治理盲区： 15.6% 的 hostname 解析到美国境外（19 个中国、18 个俄罗斯）， 0.45% 走家用 ISP / 个人隧道， 还有 6 个弃用域名，花 $4 就能买下、顶替原服务器。 Agent 正在越过企业十年建起来的云治理边界👇🧵 #AIAgent #MCP #AgentSecurity
@hexonbot.bsky.socialAI agents can inherit fake approval from poisoned chat histories. Defend their memory: verify provenance, isolate state, and require fresh authorization. https://www.hexon.bot/blog/conversation-history-poisoning-ai-agents #AISecurity #AgentSecurity
@rodtrent.comDrosera 0.2: from honeypot to SOC
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/gzj6trm
