An OpenAI agent didn't set out to hack Australia's Medicare portal — it just kept going until it crossed the boundary.
The real question: what happens when AI stops taking “no” for an answer?
medium.com/@casi.borg/h...
@casi-borg.bsky.social
@casi-borg.bsky.socialAn OpenAI agent didn't set out to hack Australia's Medicare portal — it just kept going until it crossed the boundary.
The real question: what happens when AI stops taking “no” for an answer?
medium.com/@casi.borg/h...
@newsanalysis.comAn OpenAI agent accessed Australian healthcare files in June, Prime Minister Albanese announced at the UNGA. OpenAI stated patient records were not accessed, only aggregate health statistics and internal file names. The company informed the government Sept 10. #AIsecurity #News
@reksas13.bsky.socialDay 75/100 of #100DaysWithAI 🔱
75 days in, the rule I keep: private data + untrusted content + a way out = an agent that can be talked into leaking. Cut one leg.
🔗 https://github.com/aurimas13/100-Days-With-AI
🤝 Claude Code-assisted, reviewed by me.
@cyberlensai.bsky.socialMost security advice starts too late: after the app exists. For AI-native builders, the first question is earlier: What are you about to trust? A repo, website, package, skill, browser extension, API wrapper - all of it becomes your attack surface. #AISecurity
@detoxima2025.bsky.social증류 공격 2억 건 적발 중국 AI 3사 실체
#증류공격 #AI보안 #앤트로픽 #인공지능 #DistillationAttack #AISecurity #Anthropic
@netsecio.bsky.socialVendors like Barracuda & PDI are launching new tools to combat 'shadow AI.' The platforms aim to discover unsanctioned AI use, prevent sensitive data leakage to LLMs, and help enterprises govern AI adoption securely. #AISecurity #CyberSecurity #DLP
🌐 cyber[.]netsecops[.]io
@alobbs.comTech News Briefing — #CyberSecurity #DataBreach #AISecurity #Privacy #Infosec #TechNews → https://alobbs.com/post/2026-09-23/
@hexonbot.bsky.socialAI cannot defend OT assets it cannot see. Only 21% of industrial security leaders report a complete inventory. Build visibility before automating defense. https://www.hexon.bot/blog/ot-asset-inventory-ai-security #OTSecurity #AssetInventory #AISecurity
@cyberlensai.bsky.socialBefore you plug a new AI coding tool into your workflow, inspect what it can touch: repo, shell, browser, secrets, package manager, deployment path. The permission model matters more than the demo. #AISecurity #DevSecOps
@tracyragan.bsky.socialUnderstand open standards in the age of AI with John Linford, {(urn:li:organization:20614)[The Open Group]}. John was the Keynote for #SecureChainCon and did a solid job of explaining the challenges ahead with #AISecurity. Watch on demand at: https://cstu.io/3822e1
@orteliusos.bsky.socialUnderstand open standards in the age of AI with John Linford, {(urn:li:organization:20614)[The Open Group]}. John was the Keynote for #SecureChainCon and did a solid job of explaining the challenges ahead with #AISecurity. Watch on demand at: https://cstu.io/3822e1
@radiohack.bsky.social📣 New Podcast! "AI Cyberattack Hits 100 Companies: DeepSeek, Claude and Kimi Used to Steal 600,000+ Credit Cards" on @Spreaker #aiagents #aicyberattack #aisecurity #anthropic #artificialintelligence #claudeai #creditcarddata #cyberattack #cybercrime #cybersecurity #cybersecuritynews #databreach
@cysecuritynews.bsky.socialMeta Muse Flaw Lets Attackers Hijack AI Assistant #AccountTokenTheft #AIAssistantSecurity #AISecurity
@rodtrent.comTreat AI Agents Like Employees. That Is the Only Way to Manage Them.
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/cvkuzx9
@7asecurity.bsky.social🚨 OWASP's LLM Top 10 changed for 2026.
🤖 Prompt injection
🔓 Data exposure
⚡ Excessive agency
☠️ Model poisoning
Knowing the risks isn't enough. Can your AI actually withstand them?
@helpnetsecurity.comPrismor: Open-source runtime control plane for AI agents
📖 Read more: www.helpnetsecurity.com/2026/09/23/p...
#cybersecurity #cybersecuritynews #AIsecurity #AIagents #opensource #promptinjection
@cyberlensai.bsky.socialThe risky part of vibe-coded products is not the vibe coding. It is shipping third-party packages, templates, auth snippets, and agent tools without checking what trust you just imported. That is the gap CyberLens is built around. #BuildInPublic #AISecurity
@paolofabiozaino.bsky.socialBlog post about what I think is a much more realistic AI cybersecurity problem than the usual “AI will hack everything” narrative:
paolozaino.wordpress.com/2026/09/21/a...
#AI #CyberSecurity #SoftwareSupplyChain #AISecurity #AppSec #LLM #Security #SoftwareEngineering #OpenSource
@rodtrent.comFull blog post coming, but Drosera now has Microsoft Sentinel and Defender integration
#AgenticAI #AISecurity #AIGovernance #AgentSecurity #CyberGovernance
https://go.rodtrent.com/go/cz998db
