HydroX AI

🚨 Learn Red-Teaming with HydroX AI & HumaneIntelligence at #RSA2025 🚨 Don't miss our RSAC learning lab, "Human vs Machine," on April 28th! You'll get hands-on experience uncovering vulnerabilities in LLMs in a collaborative environment. Led by Victor Bian (COO, HydroX AI) and Theodora Skeadas (Chief of Staff, Humane Intelligence), this session will cover: 🔑 Adversarial testing workflows 🔐 Red-teaming hands-on practices 🤖 Strengthening AI security for the future This is a must-attend event for anyone involved in or concerned about the security of AI, particularly LLMs. 🔗 Learn more here: https://www.hydrox.ai/rsac #RSA2025 #AIsecurity #redteaming #LLM #HydroXAI #HumaneIntelligence #AISafety #AI #MachineLearning

🚀 From Research to Reality: HydroX AI at #ICLR 2025 We’re thrilled to announce that Zhuo Li, CEO & Founder of HydroX AI, and Xuying Li, AI Engineer at HydroX AI, will be speaking at the ICLR 2025 Workshop on Human-AI Coevolution in Singapore! 📅 Date: April 27, 2025 📍 Location: Singapore EXPO 🔗 Register now: iclr.cc/Register2 ✅ Zhuo Li will share cutting-edge insights on AI safety and risk mitigation, bridging theory with real-world applications from HydroX AI’s pioneering work. ✅ Xuying Li will explore LLM safety, adversarial robustness, and AI governance, showcasing her research on building resilient and ethical AI systems. ✨ Join us in Singapore for an exciting discussion on the coevolution of humans and AI, and hear from two of the most insightful minds working on the frontier of safe and ethical AI! #ICLR2025 #AI #HumanAICoevolution #ResponsibleAI #AIResearch

🎤 Exciting news! HydroX AI COO, Victor Bian, will be speaking at the upcoming #AIA Talk on: AI Red-Teaming — Stress Testing AI Systems for Safety & Reliability 📅 Date: April 17, 2025 | 🕘 9:00–10:00 AM GMT-7 🔗 Join the talk: https://lnkd.in/gQ8G6hvT Victor will dive into practical red-teaming strategies, covering topics such as prompt injection and harmful content detection. He’ll also share valuable insights from HydroX AI’s platform and our collaborations with leading partners like Anthropic. Don’t miss this opportunity to learn from an expert in the field! Looking forward to seeing you there! #AI #RedTeaming #AIsecurity #HydroXAI #AIAI #TechTalk #AIreliability

Had a great time meeting amazing founders and the Google team at #NEXT! Thanks for having us!

🎉 We’re excited to share that HydroX AI’s COO, Victor Bian, took the stage today at #GoogleCloudNext2025 to deliver our official company pitch! 🎤 In his presentation, Victor showcased how the HydroX AI platform is redefining AI security and shared our bold vision for a safer, more trustworthy AI future. This moment marks our first step as part of the #GoogleAccelerator, and we’re looking forward to sharing more insights and innovations in the coming weeks! #GoogleCloudNext #AI #GoogleAccelerator #Cybersecurity #AITrust #GenAI

🚀 Building Accessible & Effective Secure AI: Our Exciting Start at Google Cloud Next! This week is a game-changer for HydroX AI! We are beyond excited to announce that we are officially part of the 2025 cohort of the Google for Startups Cloud AI Accelerator — a prestigious 10-week program designed to support North American AI startups with mentorship, technical support, and access to Google’s powerful AI tools. What makes this even more special? We’re launching this incredible journey live this week at #GoogleCloudNext2025 in Las Vegas! 🌟 This is just the beginning of an amazing opportunity to collaborate with Google’s teams and other top startups, and accelerate our mission to create secure, scalable AI solutions. Over the next few months, we’ll be pushing boundaries, refining our product, and expanding our impact on AI security. We can’t wait to share the journey ahead and are so grateful to be part of this incredible program. 👉 Let’s connect if you’re focused on building secure AI for the future! #GoogleForStartups #AcceleratedWithGoogle #GoogleCloudNext #Startups #Innovation #CloudNEXT25

🚀 New Paper Release: Optimizing Safe and Aligned AI with GRPO 🤖 At HydroX AI, we introduce Group Relative Policy Optimization (GRPO) with Multi-Label Reward Regression, a more efficient, scalable, and interpretable method for safe and aligned language models. 🔍 Why GRPO? GRPO compares output groups to optimize policies, eliminating the need for a value critic. Our multi-label reward model evaluates multiple alignment aspects (e.g., safety, helpfulness) for a balanced training process. 📝 Key Findings Higher Alignment: Improves safety and truthfulness across all LLM sizes (0.5B-14B parameters). Lower Cost: GRPO is computationally efficient, achieving better alignment with fewer resources. More Control: Dynamic rebalancing of alignment priorities without retraining. 💡 Future Work Scale diverse datasets, improve human oversight, and extend to code generation and autonomous agents. 🤖 Check the Paper here: https://lnkd.in/gXg-XgGH ✍️ By Xuying Li, Zhuo Li, Yuji Kosuga, Victor Bian from HydroX AI We’re committed to open research and will release all models on Hugging Face. Let’s shape the future of safe, aligned AI! 💪 #AI #MachineLearning #AIAlignment #HydroXAI #SafeAI #Innovation #FutureOfAI

🙌 Humane Intelligence at RSAC 2025: HUMAN VS. MACHINE – Exploring the Future of AI Security We’re thrilled to join forces with HydroX AI to present HUMAN VS. MACHINE at the RSA Conference 2025! This high-impact session will push the boundaries of AI security through cutting-edge adversarial testing designed to expose vulnerabilities and fortify defenses. - What You’ll Experience: Dive into live simulations with experts, researchers, and AI professionals as they put security systems to the test. These hands-on exercises aim to refine strategies and shape a more secure and resilient AI future. - Why Attend: As AI continues to evolve rapidly, so do its risks. This session offers a rare opportunity to drive innovation and deepen cross-sector collaboration—spanning tech, healthcare, finance, and defense—to safeguard the future of AI. Join us at #RSAC2025 and be part of the conversation that’s defining the next era of AI security. 🔗 Learn more: https://rsac.hydrox.ai/

Excited to announce that I’ll be speaking at HydroX AI’s HUMAN VS. MACHINE session at RSA Conference 2025! 🤖 This isn’t just another session for me—it's a passion project where we dive deep into AI security through innovative adversarial testing. I can’t wait to share my insights and join forces with experts, researchers, and AI enthusiasts to uncover vulnerabilities and strengthen our defenses. 🔐 What’s in Store: I’ll be leading alongside Theodora Skeadas and Dr. Rumman Chowdhury immersive simulations designed to help stakeholders understand the risks of AI systems. Together, we'll refine our strategies and collaborate on building a safer, more resilient AI future—a mission that’s close to my heart given the rapid evolution of AI. 🔍 Why It Matters to Me: Securing AI is more than a technical challenge; it’s about safeguarding innovation across industries like tech, healthcare, finance, and defense. I’m excited to help spark conversations that drive meaningful change and contribute to a future where AI benefits us all. Excited to see y'all soon~ 🔗 Stay tuned for more details: https://www.hydrox.ai/rsac #RSAC #ApplicationSecurity #Cybersecurity #OnlineSafety #AI #ArtificialIntelligence #Innovation

🚀 HydroX AI at RSAC 2025: HUMAN VS. MACHINE – Exploring the Future of AI Security 🤖 We’re excited to bring HUMAN VS. MACHINE to #RSAC Conference 2025! This cutting-edge session will delve into AI security, focusing on adversarial testing to uncover vulnerabilities and strengthen defenses. 🔐 What to Expect: Experts, researchers, and AI professionals will participate in immersive simulations to test security tactics and refine defense strategies, helping shape a safer and more resilient AI future. 🔍 Why It Matters: With the rapid evolution of AI, securing it is more critical than ever. This event provides a unique opportunity to exchange knowledge, drive innovation, and explore AI security in sectors like tech, healthcare, finance, and defense. 🚀 Join us at RSAC 2025 to learn, collaborate, and contribute to the future of AI security! 🔗 Stay tuned: https://www.hydrox.ai/rsac RSA Conference #RSAC #applicationsecurity #cybersecurity #OnlineSafety #AI #artificialintelligence #Innovation