SonnyLabs.ai

Wrapping up an incredible journey with NovaUCD’s AI Accelerator! Here’s a snapshot from our final pitch. A huge thank you to our panel—Saurabh Kumar , Niamh Sterling and Roopesh Bharatwaj KR for their invaluable feedback and guidance throughout. We’re excited for what lies ahead—onward and upward from here! #AIEcosystem #Innovation #SonnyLabs #Growth

A huge thank you to NovaUCD, CeADAR Ireland, and Enterprise Ireland for guiding and supporting SonnyLabs.ai and the incredible cohort of startups in UCD’s inaugural AI Ecosystem Accelerator Programme. This 6-month journey has been invaluable, offering us the tools, insights, and connections to bring our AI solutions to a global stage. Congratulations to our fellow participants: ALPACA Assessment, Customs Window, Durotimi AI, Frontier AI, GoMappED Limited, Knowcarbon, Vox Med AI, La Casa and Wrksense—it's been inspiring to grow alongside each of you. Thank you to Professor Kate Robson Brown and the entire NovaUCD and CeADAR teams for creating an impactful program that’s shaping the future of AI in Ireland. We’re excited to continue this journey and make a positive impact!

We’re really excited to share that SonnyLabs.ai has been invited to join NDRC’s Masterclass, Harnessing Generative AI: A Practical Guide for the Startup Ecosystem. Our founder, Liana Tomescu, will be on the Gen AI Founder Panel, talking about the latest trends, real-world applications, and what’s changing in the venture landscape. It’s awesome to be alongside visionary founders like Garry Tiscovschi (Kreoh), Marie Toft(Emotionise AI), Hrvoje Smolic 🔮 Smolic (Graphite Note), and Óran McKenna O'Flynn (Talio). Plus, we’re looking forward to hearing from VCs like Mairéad Roche (Act Venture Capital), Jeremy Debono (Elkstone), and Dylan Scully (Frontline Ventures).

As AI adoption accelerates across industries, AI-SPM is becoming a critical component of enterprise cybersecurity. With the increasing use of managed AI services and SDKs, businesses are facing new, complex risks that traditional cybersecurity solutions can't address. AI-SPM helps secure AI models, pipelines, data, and services by offering: Comprehensive Inventory Management: Understand what AI assets your enterprise is using and their security risks. Full-stack Visibility: Gain complete visibility over both self-hosted and managed AI services without compromising performance. Training Data Security: Protect training data from attacks, such as AI poisoning or data exfiltration. Proactive Risk Management: Identify attack paths early to mitigate risks before they become large-scale issues. AI Configuration Rules: Establish security baselines and detect misconfigurations, like exposed IPs and endpoints. With generative AI fueling $4.4 trillion in potential global economic impact, securing AI is no longer optional—it's essential. AI-SPM, standing for AI- Security Posture Manasgement, is a vital part of any cybersecurity stack, ensuring that enterprises can safely adopt AI technologies while protecting against privacy breaches, enhanced cyberattacks, misinformation, and identity fraud. Try SonnyLabs to see how to protect your GenAI systems now.

Congratulations to Skippio and Prepsheets on this milestone! Happy to see our colleagues at NovaUCD grow !

At the forefront of AI advancements, LLM platforms like ChatGPT are expanding with third-party plugins that enhance functionality but introduce unique security, privacy, and safety challenges. A recent study led by Umar Iqbal, Tadayoshi Kohno, and Franziska Roesner presents a comprehensive evaluation framework, focusing on: Attack Taxonomy: Outlining potential risks and vulnerabilities in LLM plugin ecosystems. Case Analysis: Applying the framework to OpenAI’s plugin ecosystem, uncovering real security risks. Future Recommendations: Insights for securing emerging LLM platforms and anticipating the next wave of AI-integrated systems. To read the full article: https://lnkd.in/dWqzBNJg #AI #Cybersecurity #LLMPlatforms #Innovation #DataPrivacy #TechResearch

Last week, SonnyLabs.ai had the chance to join 'The Dublin Startup Scene – Networking event, hosted by Samuel Klacman, Quinn Bonta and volunteer Anusha Makan with the fantastic team at Rare Founders. It was an energetic evening filled with networking, pitching, and connecting with over 100 founders, investors, and movers in the Dublin tech scene. The open-mic pitching format was an exciting way to see fellow founders pitch their ideas, and it’s always inspiring to witness the ambition and innovation driving Ireland's startup ecosystem. Huge thanks to Sam and Rare Founders for creating such an electric environment – definitely a night to remember! 🚀

With the integration of AI across sectors like national security, critical infrastructure, and healthcare, the need to secure machine learning systems has never been greater. A recent report by Dr. Sven Herpig sheds light on the vulnerabilities AI systems face, particularly regarding the attack surfaces exposed by machine learning. Key Highlights: Training Data Vulnerability: Attackers can manipulate training data, influencing machine learning models with methods like data poisoning. Deployment and Input Risks: Input processing stages, such as those in autonomous systems, are susceptible to adversarial attacks, potentially leading to misclassification and incorrect actions. Supply Chain Security: The security of AI's complex supply chain—training data, pre-trained models, and fine-tuning—requires rigorous protection and transparency. Read the full article here: https://lnkd.in/d6KwU_JD

As the adoption of large language models (LLMs) skyrockets across industries like education and healthcare, it’s crucial to address their vulnerabilities. Recently, a working group, featuring experts from various prestigious institutions, explored the security implications of LLMs, particularly their susceptibility to adversarial attacks. Key takeaways from the paper: Vulnerability Analysis: LLMs, unlike traditional ML models, face unique risks, including issues related to hallucinations and data poisoning. Complexity of Risk Assessment: Evaluating the security of LLMs involves navigating opaque algorithms, diverse applications, and rapid technological advancements. Supply Chain Challenges: Understanding the data flow—from training to fine-tuning—is essential for identifying vulnerabilities and ensuring model integrity. At SonnyLabs, we are committed to innovating in cybersecurity solutions tailored for AI applications, ensuring that your interactions with LLMs are secure and trustworthy. Read the full paper here: https://lnkd.in/dipDcmYh

SonnyLabs.ai had a fantastic time attending Startup Day, powered by Dogpatch Labs and Google, packed with networking, mentoring, and insights, bringing together Ireland's vibrant startup community. Whether you’re just launching or scaling your business, this event offered invaluable resources and support for startups. It was also inspiring to hear the entrepreneurial journeys of Matthew Coffey from SQUID, Karl O'Brien 🦸 from StoreHero , and Greta Dunne from Marker Video – their stories are a true testament to the innovation happening in Ireland! A big thank you to Samuel Klacman, Clíona Jordan ☘ , and the team at Google for Startups for hosting us, and to Dogpatch Labs and NDRC for making it all possible. Looking forward to applying these insights as we continue to grow at SonnyLabs.ai! 🚀 #SonnyLabs #StartupDay #Google #StartupCommunity #Innovation #IrishStartups #DogpatchLabs #NDRC #Networking #Mentorship #TechEvents