OASIS的动态

?? Exciting Partnership with #??????????, an OASIS Open Project!?? CipherData is proud to announce our membership in the ?????????????????? ?????? ???????????? ?????(#CoSAI), standing alongside industry leaders like Google, OpenAI, Microsoft, Anthropic, Amazon, along with 32 other organizations dedicated to advancing AI security. We’re also thrilled to share that our CEO, Kevin Ha, has been appointed to the ?????????? ?????????????? ?????????????????? ??????????, where he’ll bring his expertise and vision to help shape the future of AI security. CoSAI’s mission — to democratize AI #security and #privacy techniques, expertise, and solutions — aligns with CipherData’s core commitment: safeguarding our clients through cutting-edge, responsible AI technology. As part of #CoSAI, #CipherData will not only contribute to industry-wide standards but will also integrate these latest AI security innovations into our AI-Driven #SecurityOperations platform, ensuring continuous protection and risk management for our customers. Together, we’re working toward "???????????? ???? ?????????????? ???????????? ?????? ??????". ?? Learn more about CoSAI’s initiatives: https://lnkd.in/ejBSa8j3 #CipherData #CoSAI #OASIS #AI #SecOps #cybersecurity #AISecurity #partnerships #opensource #openstandards

?? Innovation vs. Security: The DeepSeek Dilemma ?? The AI landscape is evolving rapidly, and DeepSeek is making waves! ???? With its R1 model, DeepSeek has managed to achieve performance levels comparable to GPT-4, but at a fraction of the cost—just $6 million vs. OpenAI’s $100 million+. How? By leveraging the Mixture of Experts (MoE) technique, optimizing computational efficiency like never before. ?? This game-changing AI breakthrough has already shaken global markets, contributing to a $593 billion drop in Nvidia’s valuation and affecting major tech giants like Microsoft and Google. ?? DeepSeek’s open-source approach is democratizing AI development, pushing the boundaries of competition and innovation. ?? Even with U.S. sanctions, DeepSeek has thrived using less advanced chips, proving that AI progress isn’t always dependent on cutting-edge hardware. However, there’s a catch—security. ???? Recently, DeepSeek suffered a major data breach, exposing over a million lines of chat logs, API keys, and backend metadata through an unsecured ClickHouse database. This breach highlights an uncomfortable reality: No matter how powerful AI is, poor security can destroy its credibility. ?? The Cost of Negligence in Cybersecurity: ? Unsecured databases → Unauthorized access, leading to potential data theft. ? Exposed API keys → Attackers gaining control over critical systems. ? Escalated threats → Hackers exploiting weaknesses for financial and reputational damage. ?? Key Takeaway: AI leadership isn’t just about innovation—it’s about security. Organizations must implement strict access controls, encrypted data storage, and real-time monitoring to prevent such disasters. DeepSeek’s rise proves that the AI race is more intense than ever, but its security lapse is a cautionary tale. As AI dominates, cybersecurity must evolve alongside it. Would you trust an AI system that isn’t secure? ???? Let’s discuss in the comments! ?? #AI #Cybersecurity #DeepSeek #TechInnovation #DataSecurity #FutureOfAI #ArtificialIntelligence #SOC #CyberThreats #MachineLearning

#AI & #Cybersecurity- Key insights from Access Partnership report ? as AI is increasingly integrated into critical #systems and #infrastructures, it introduces significant #cybersecurityrisks that must be effectively managed ? While AI holds the potential to deliver #transformative advancements, it also brings unique #vulnerabilities, such as #datapoisoning, model evasion attacks, and #ethical concerns, which could compromise the #integrity and #security of #AIsystems. ? As national #cybersecurity strategies, establish AI-specific guidelines, invest in #research and development, foster international #cooperation, and promoting #AIliteracy are crucial to ensure that the deployment of AI systems is both #secure and #ethical, supporting innovation while proactively addressing #emergingthreats. #Bigdata #Artificialintelligence #Dataprivacy #Databreach #dataprotection #security #Digitalidentity #Fintech #Finserv #Regulation #RegTech #Digitalidentity #Security Mike Flache Francesco Burelli Tony Moroney Panagiotis Kriaris Dr. Martha Boeckenfeld Prof. Dr. Ingrid Vasiliu-Feltes Spiros Margaris Alex Jimenez Nicolas Babin Nicolas Pinto Amitav Bhattacharjee Sam Boboev Efi Pylarinou Imtiaz Adam Sally Eaves Eveline Ruehlin Martin Moeller Dr. Debashis Dutta PhD Machine Learning Speciality Solution Architect (Google/AWS) PMP? PMI-RMP?, SAS https://lnkd.in/gF5C3dWx

?? The Rise and Crash of DeepSeek: A Wake-Up Call for Tech Just days ago, DeepSeek was the talk of the AI world: ?? Breaking OpenAI’s download records ? Promising GPT-4-level performance at unbeatable prices Then came the unexpected collapse. ?? A massive cyberattack ?? Registrations frozen ?? Potential data breach ??? Security teams scrambling But here’s the part no one’s talking about: This isn’t just a security issue — it’s a lesson for the entire tech industry. ?? When growth becomes reckless ?? When corners are cut ?? When giants are challenged without proper foundations The cracks start to show. The same mistakes have been seen time and time again. ?? Cheap solutions without thorough vetting ?? Flimsy security protocols ? Rushed releases What follows is predictable: ?? Security vulnerabilities ?? Hungry hackers ?? Burned users DeepSeek is a reminder that in tech, speed isn’t everything. Success requires more than innovation. It demands trust, security, and a commitment to doing things right. The AI race is heating up, but it won’t be won by those who cut corners. DeepSeek may recover, but trust is hard to rebuild once it's lost. ?? The future belongs to those who prioritize stability, security, and excellence over shortcuts. PS: How secure are your systems today? ?? #AI #TechNews #Cybersecurity #DeepSeek #Innovation #Trust #TechInsights #DataSecurity #LessonsLearned #ArtificialIntelligence

???????????????? ?????????? ?????? ?????????????? ?????????????????? ?????????????? ???? ?? ?????????? ???????? ???????????? Chinese AI startup DeepSeek left critical user data and internal secrets unprotected, raising alarms over security risks in the AI industry. Over 1 million lines of sensitive data—including user chat histories, API secrets, backend details, and even plaintext passwords—were left wide open in an unsecured database. ?????? ?????????? ????????? It took less than an hour to fix… but how long was it exposed? ?? ?????? ???????????? ???????????????????? ?? ???????????????? ??????????????:? AI innovation is sprinting ahead of security. As companies race to deploy cutting-edge models like DeepSeek’s “cost-effective” reasoning AI, basic misconfigurations are leaving gaping holes for attackers. ?? ?????????? ???????? ??????’?? ??????: “This was so simple to find, we believe we’re not the only ones who found it.” With global regulators (from the White House to Italy and Ireland) now scrutinizing DeepSeek’s data practices, the incident raises urgent questions: Can enterprises trust third-party AI tools with their data? Are we sacrificing security for speed in the AI arms race? Is your organization prioritizing “smart” over “safe”? As AI reshapes industries, this breach is a wake-up call: Innovation without security is a ticking time bomb. ?? What’s your take? ?? Can AI startups balance breakneck growth with ironclad security? ?? Would this incident make you rethink adopting emerging AI solutions? Let’s discuss below. ?? Image Credit: WizResearch #AISecurity #Cybersecurity #DataPrivacy #TechNews #AIInnovation #CyberRisk #DeepSeek #WizResearch #TechEthics

Startup Spotlight: Knostic Tackles AI's Oversharing Problem with Innovative Cybersecurity Solutions Knostic, a cutting-edge cybersecurity startup, has emerged as a finalist in the prestigious Black Hat USA Startup Spotlight competition, thanks to its pioneering approach to addressing AI's oversharing problem. As AI tools, especially generative AI, become increasingly integrated into enterprise environments, the risk of sensitive data being inadvertently shared or leaked has escalated. Knostic's technology is designed to provide visibility, control, and curation over the data handled by generative AI systems. By implementing robust guardrails, the startup ensures that enterprise data remains secure, preventing accidental leaks or unauthorized access. Their initial focus is on safeguarding Microsoft's Copilot, a powerful AI assistant integrated into various Microsoft tools. However, Knostic's vision extends far beyond this, with plans to expand their solutions to other SaaS platforms that incorporate large language models (LLMs). In an era where data privacy and security are paramount, Knostic's innovative approach to managing AI's data handling is crucial for enterprises looking to harness the power of AI without compromising on security. #Cybersecurity #AI #StartupSpotlight #BlackHatUSA #DataPrivacy #GenerativeAI #MicrosoftCopilot #Knostic #EnterpriseSecurity

???Exciting News from Skyflow! Skyflow has just unveiled its?Agentic AI Security and Privacy Layer, empowering enterprises to build and deploy AI agents with robust security and privacy features. ??? ???Why Agentic AI Needs a Security and Privacy Solution: -Sensitive Data Access: AI agents often handle critical information like financial data, healthcare records, and intellectual property. Without strong security measures, this data is vulnerable to exposure or misuse. ?????? -Compliance Challenges: Managing sensitive data requires adherence to regulations such as GDPR, HIPAA, and CCPA. Non-compliance can lead to legal issues and erode user trust. ????? -User Trust and Reputation: Data breaches can severely damage an organization's reputation. Implementing privacy-preserving measures is essential to build and maintain user trust. ??? ???How Skyflow's Agentic AI Security and Privacy Layer Helps: -Data Privacy Vault: Encrypts, tokenizes, or de-identifies sensitive information during model training, preventing private data from being embedded in AI models and reducing exposure risk. ?? -Secure Tool Interactions: Acts as a gatekeeper for external tool and API integrations, ensuring only de-identified or masked data is accessed, safeguarding against unauthorized exposure. ???? -Granular Access Controls and Auditing: Provides fine-grained access controls, data masking, and redaction, along with comprehensive logging and audit trails to ensure compliance and maintain accountability. ??? ?? We're at?AWS re:Invent?this week! Stop by?booth #357?to learn more about how Skyflow is revolutionizing AI security and privacy. Come say hi—we can't wait to share more with you! ???? Or message me if interested in meeting with our team to learn more! #AWSreInvent #AI #DataPrivacy #Cybersecurity #Innovation #Skyflow #PrivacyFirst ??

Great insights from Battery Ventures on the "State of Enterprise Tech Spending" Check it out --> https://lnkd.in/gSbjvvsX One key insight in the report is that the vast majority of buyers (84%) plan to increase spending on AI/ML tools. Also, 89% of organizations leveraging that have LLMs in production are utilizing multiple tools to improve model performance (use case performance)... Example included tools for LLM observability/evaluation (64% ) and model security (55%). I'm biased as the CEO of Protect AI, but there should be no AI in the enterprise without security of AI. The Battery report echoes what we are seeing in the market. Yes, AI is in production with more use cases / deployment planned across all industries and business units. Yes, enterprises understand that they must have trusted, safe, and secure AI. Spend for AI continues to increase, along with budgets to safeguard AI. Happy to continue the conversation here or in the MLSecOps Community. I'm going to part part in an "ask me anything" session on MLSecOps.com in June - see you here! Nice work Battery! Evan Witte Danel Dayan Scott Goering Jason Mendel Patrick Hsu #protectai #mlsecops #aisecurity #artificialintelligence #cybersecurity

The success of AI deployments hinges on the quality, security and integrity of the data used to train and operate these systems. Data governance plays a vital role in ensuring that AI systems are built on a foundation of trust and reliability. Data governance is no longer an optional component of AI development. It is an essential foundation for ensuring the reliability, security, and ethical use of AI systems. By embracing best practices for data governance, organizations can unlock the full potential of AI and generative AI while mitigating risks and fostering trust. As AI continues to evolve, data governance will remain a critical element for navigating the complexities of this rapidly advancing technology. ? #DataGovernance #AI #GenerativeAI #DataManagement #DataQuality #Compliance #CyberSecurity #Innovation

?? Data: The Lifeblood of AI Innovation ?? In today’s world, data is more than just information—it's the fuel that powers businesses and drives groundbreaking AI applications. Much like a living organism, data is constantly being generated, enhanced, evolved, and repurposed. ???? But here’s the challenge: as data grows, so do the complexities in managing it. How do we navigate the ever-changing landscape of data regulations and requirements while still pushing AI innovation forward? ?? It’s time to explore pragmatic approaches that balance data ethics, privacy, and AI development. Whether you're a startup or an industry leader, adapting to the future of data is critical to staying ahead. Let’s shape the future of AI—responsibly and strategically. #AI #DataInnovation #MachineLearning #AIethics #TechTrends #Cybersecurity #AIforBusiness