ᅠᅠ ᅠᅠ ᅠᅠ kanekoᅠᅠshimada.ᅠᅠ[ ᅠᅠinhumanᅠᅠ] ᅠᅠ energyᅠᅠ transaction ᅠᅠ ᅠᅠ japanese ᅠᅠ exᅠ¡ᅠᅠ𝚂𝙷𝙸𝙴𝙻𝙳ᅠᅠagentᅠᅠ*ᅠᅠlevelᅠ 1 ᅠᅠ

🥉 3rd Prize — Musical Equity, ShieldAgent & Ebikatsu Musical Equity pairs indigenous music with a persistent AI agent that carries cultural context, usage permissions, and licensing rules—ensuring music spreads with understanding, not anonymity. ShieldAgent is an AI agent system that analyzes work contracts, detects labor law violations, and preserves tamper-proof evidence to help protect migrant workers in Japan. Ebikatsu: Honor Revolution by PHR Onchain, exploring on-chain narratives around honor, identity, and community.

Agentic AI is scaling fast. But safety? Not fast enough @recallnet X @build_on_bob New research tells two sides of the story ShieldAgent hits ~90% recall by embedding verifiable policies directly into agent logic. A real leap in enforceable safety. But then comes Safety Devolution agents with tool access (search, plugins, retrieval) grow more useful… …and more dangerous. Less refusal. More hallucination. Harder to trust. This isn’t theoretical. These agents are already inside crypto, support systems, games, governance. And that’s why BOB matters. With ZK fraud proofs, a live BitVM bridge testnet, and native Bitcoin security, BOB isn’t just scaling it’s building a safe, verifiable execution layer for agents. Not wrapped BTC. Not vaporware. Just programmable Bitcoin, with cryptographic finality. Because the future isn’t just autonomous. It must be provably safe. Build on Bitcoin and build responsibly.

Wataa 💋 #mariahill #shieldagent

VirtueGuard is now on Together AI! We’re excited to bring enterprise-grade AI security & compliance Guardrail solution to Together’s powerful, developer-friendly platform. @togethercompute makes it simple to build and scale AI in production, and now you can add VirtueGuard’s protection with just a single API parameter. What makes VirtueGuard stand out? ⚡ Under 10ms 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗲 - faster than any alternative guardrail solution 🎯 𝟴𝟵% 𝗮𝗰𝗰𝘂𝗿𝗮𝗰𝘆 vs 76% (AWS Bedrock) 🧠 𝗖𝗼𝗻𝘁𝗲𝘅𝘁-𝗮𝘄𝗮𝗿𝗲 - adapts to your policies, not just keywords 🔄 𝗔𝘂𝘁𝗼-𝘂𝗽𝗱𝗮𝘁𝗶𝗻𝗴 broad risk coverage, including special advice, jailbreaks, prompt injections, etc. VirtueGuard delivers comprehensive coverage across diverse risk categories, with industry-leading speed, accuracy, and customization, now available on Together’s reliable infrastructure. Try it today → together.ai/blog/virtueguard Related research from Virtue AI on guardrail solutions: - RigorLLM: Resilient Guardrails for Large Language Models against Undesired Content (cited by Llama Guard). arxiv.org/pdf/2403.13031.pdf - GuardAgent: Safeguard LLM Agents by a Guard Agent via Knowledge-Enabled Reasoning (First Guardrail solution for agents). arxiv.org/abs/2406.09187 - ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning (Guardrail solution for agentic systems with guarantees). arxiv.org/abs/2503.22738 #AIsecurity #responsibleAI #ArtificialIntelligence #AI #MachineLearning #LLM #AIagent #AgenticAI #VoiceAI #CodeGen #AIstrategy

Wataa 💋 #mariahill #shieldagent

Wataa 💋 #mariahill #shieldagent

Sparking AI Security Research Impact at ICML 2025: We’re thrilled to share that the Virtue AI team had 19 papers accepted at ICML 2025, one of the most prestigious conferences in AI and machine learning. Our work spanned critical advances in Security for AI agents and AI systems, including: verifiable safety reasoning, knowledge-enabled guardrails, dynamic red teaming, provable defenses, and more. These contributions reflect the passion and depth of our team, leading researchers and practitioners in AI security,  who are pushing the frontier of the science and practice of trustworthy and secure AI. Missed us at ICML? We’d love to connect! Reach out if you’re interested in learning more about how we’re building the trust layer for the next generation of intelligent systems. 1. ShieldAgent: Shielding LLM Agents via Verifiable Safety Policy Reasoning 2. GuardAgent: Safeguard LLM Agents via Knowledge-Enabled Reasoning 3. UDora: A Unified Red Teaming Framework Against LLM Agents by Dynamically Leveraging Their Own Reasoning 4, AdvAgent: Controllable Blackbox Red-teaming on Web Agents 5. SafeAuto: Knowledge-Enhanced Safe Autonomous Driving with Multimodal Foundation Models 6. Reliable and Efficient Amortized Model-based Evaluation 7. Underestimated Privacy Risks for Minority Populations in Large Language Model Unlearning 8. [Provable Agent Security] MELON: Provable Defense Against Indirect Prompt Injection Attacks in AI Agents 9. [Agentic System for Security] PatchPilot: A Cost-Efficient Software Engineering Agent with Early Attempts on Formal Verification 10. How do large language monkeys get their power (laws)? 11. Putnam-axiom: A functional and static benchmark for measuring higher level mathematical reasoning 12. Certified unlearning for neural networks 13. Collapse or thrive? perils and promises of synthetic data in a self-generating world 14. Can large language models ask the right questions in solving complex problems with incomplete information? 15. Learning to (learn at test time): RNNs with expressive hidden states 16. Logits are all we need to adapt closed models 17. Language models may verbatim complete text they were not explicitly trained on 18. Why Has Predicting Downstream Capabilities of Frontier AI Models with Scale Remained Elusive? 19. Boosting Adversarial Robustness with CLAT: Criticality-Leveraged Adversarial Training What we are building: #AIsecurity #responsibleAI for #ArtificialIntelligence #AI #MachineLearning #LLM #AIagent #AgenticAI #VoiceAI #CodeGen #AIstrategy

ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning. arxiv.org/abs/2503.22738

@PFF_NateJahnke at the top of the banana in Donkey Kong. @AndrewHallFF achieved 1.21 gigawatts in Dr. Emmett Brown. Cheech & Chong has @RyNoonan smoking. @FF_Hep casting some fireballs in Mario. SHIELDAgent has a ringer apparently in @MatthewBerryTMR (also worth a shot!)

#SFB14 I'm back in and proud to help out #FantasyCares and @ScottFish24! @MatthewBerryTMR and the #ShieldAgent division I'm coming for you!

This could not have come at a more perfect time, I love the Hammer Bros 🥹 Thanks for sharing! Excited to try this out.

Nice value with WWBN

#MarvelSnap Infinite Decks of the Week – The Fundamentals of Climbing the Ranked Ladder (Late in the Season)! ⚔️Article by @den_CCG: - marvelsnapzone.com/infinite-… Decks from: - @ShieldAgent - @Its_Anipex - @Not_Sci_Phys_HS - @wayne_206

Afaik woody had silk before and put in Jeff for the tourney since it’s easier to not clog yourself and helps vs Prof X. Also requires less thinking which adds up when you play intensely for 8 hours lol.

Gotta think there is a common denominator. #FantasyAvenger #HighestGrossingActor #JustSaying #SHIELDagent

trying something like this, but really trying to see if i can brew something with electro and mobius so i can also ramp out cards from my hand because i always get stuck with the 6-drops 😂

Dropping tomorrow on #HasbroPulse #SHIELDAgent #HYDRAAgent #2Pack #Hasbro #MarvelLegends #ActionFigures #FemaleActionFigures #ArmyBuild #FemaleArmyBuild