Giskard’s Post

🌟 Grateful to reach 10k subscribers on LinkedIn! 🌟 Thank you for being part of our journey in improving AI security, quality and compliance. As we celebrate, we're always eager to hear from you, our community: How can we make Giskard even more valuable to you and your work? 💬 Share your thoughts 🤝 Invite your network to join us 🌟 Star our GitHub repo Join us on this journey: https://gisk.ar/4eCNtg9 #AISecurity #AICompliance #community #opensource

🚀 Join us at BIG DATA & AI PARIS next week, October 15-16. 🚀 This event brings together AI and big data leaders, showcasing advancements in Generative AI and data governance. You can meet our team, Guillaume Sibout and Alex Combessie 🐢, at our booth S32 🐢 We'd love to discuss with you about AI Safety and how Giskard can help ensure the quality, security, and compliance of your GenAI applications. Join us for a live demo session, where we'll showcase how to evaluate and secure a banking chatbot using continuous LLM Red Teaming - a key step in ensuring AI safety and reliability. See you there! 🙌 📅 October, 15-16 🗺️ Paris Expo Porte de Versailles 🗣️ Demo: October 16, 11:30am Learn more and register: https://gisk.ar/4eDSRzv #BigDataAIParis #AITesting #AISecurity #GenAI #AIRedTeaming

🚀 Insights on #AISecurity from our seminar at CAPFM 📣 We recently had the opportunity to participate in the 2024 CAPFM Group's Information Systems Security Seminar. Our team, led by Guillaume Sibout and Pierre Le Jeune, presented an overview of how to test LLM-based application to ensure AI security and safety. 🚩🏴☠️ What is AI Red Teaming? AI Red Teaming is crucial for identifying vulnerabilities and mitigating risks. It helps companies address key concerns such as reputational damage, legal issues, data security breaches, and service disruptions, ensuring robust and secure LLM implementations. At Giskard, we take a holistic approach to AI testing, focusing on both security (evasion, model exfiltration, poisoning) and safety (toxicity, discriminatory content, hallucinations). We combine traditional security with responsible AI risk assessment, using both manual and automated testing methods. Learn more about AI Red Teaming 👉 https://gisk.ar/3NfhPJB #AIRedTeaming #cybersecurity #AITesting

🌟 Expanding testing to #ComputerVision tasks! 🌟 This addition to our open-source library allows you to easily detect vulnerabilities in image classification, object detection, and landmark detection applications. Data scientists can now enhance the reliability and fairness of their computer vision models. Giskard-vision allows you to: 🔹 Identify performance degradation under specific conditions or subsets of data 🔹 Detect fairness issues and biases linked to sensitive attributes 🔹Assess robustness against image perturbations like blur or noise 🔹Evaluate model performance across various image attributes such as contrast, brightness, or color 🤝 Plus, learn in our latest newsletter about our integration with NVIDIA NeMo Guardrails, bringing robust testing to LLM-based applications, and get a sneak peek at what's coming next! Read the full content here 👇 #AITesting #MachineLearning #AISafety #opensource

📣 Breaking: 116 companies, including tech giants, sign EU's #AIPact! 🇪🇺 The European Commission has released the list of AI Pact signatories, with 116 companies committing to anticipate upcoming AI regulations, who sent over 430 submissions. Major players like Amazon, Google, Microsoft, and OpenAI, along with multinationals such as Adobe, IBM, Cisco, and Samsung, are among those pledging to implement best practices. In parallel, the AI Office has announced the chairs and vice-chairs for four working groups tasked with developing the first General-Purpose AI Code of Practice. These experts, selected for their backgrounds in computer science, AI governance, and law, will guide the process from October 2024 to April 2025. As a reminder, the AI Pact is a voluntary initiative to help organizations prepare for future AI regulations. Although not legally binding, the code will provide a compliance checklist for companies, and ignoring it could elicit legal challenges. Signatories commit to making best efforts on key areas like AI governance, risk assessment, and transparency. What steps has your company taken towards AI compliance? We'd love to hear your thoughts in the comments! 👇 Official announcement 🔗 https://lnkd.in/dX69TRnD #AICompliance #AIRegulation #ResponsibleAI

📣 New article featuring Giskard in La Revue du Digital! 🎉 Following last week's AI for Finance event, and the round table on LLMs in business customer support, this article explores how BNP Paribas is leveraging AI to enhance its digital banking chatbot, with a focus on ensuring safety and reliability. As highlighted in the piece, Giskard is working closely with BNP Paribas to ensure the safety and reliability of their AI chatbot. Our platform helps identify and mitigate risks such as hallucinations, data leaks, and biased responses before deployment. Thanks to La Revue du Digital for this coverage of our work in AI safety 🙌 Read the full article here 🔗 https://gisk.ar/3BiP2kC #AISecurity #LLMeval #chatbot #AIforFinance

✨ New integration with NVIDIA #NeMo Guardrails! ✨ This integration enhances the safety and reliability of LLM-based applications. By combining Giskard with NeMo Guardrails, organizations can address critical challenges like hallucinations, prompt injection, and jailbreaks. 🚀 Developers can now: - Better detect LLM vulnerabilities - Automate rail generation  - Streamline the risk mitigation process Learn more about the integration 👉 https://lnkd.in/eBNm_sPX 🛠️ Compatible with Colang 1.0 and 2.x versions. #LLMeval #AISecurity #NVIDIA #AIGuardrails

🌐 The EU is set to launch its first AI #CodesofConduct for General-Purpose AI (GPAI)! 🌐 These codes aim to ensure that AI is ethical, safe, and aligned with European values, particularly for General-Purpose AI models. 📆 Key Dates: • September 18: The drafting process for the first AI Codes of Conduct begins, following strong interest from industry stakeholders. • September 30: The EU will hold working group discussions to further shape the content of these codes, ensuring alignment with the AI Act and international standards. For businesses, this represents a unique opportunity to shape regulations while ensuring their AI practices are aligned with European requirements. Engaging in the drafting of these codes could offer a strategic advantage in a globally connected AI ecosystem. 🔗 More information on the announcement: https://gisk.ar/3B2bU7F #AIRegulation #AISafety #EUAI #GPAI

🚀 Exciting day at AI for Finance by Artefact! ⚡️ This annual summit organized by Artefact brings together financial industry leaders, tech innovators, and AI specialists to spotlight top use cases of AI in financial services. You can still meet us at the Amazon Web Services (AWS) booth in the GenAI zone! Guillaume Sibout and Pierre Le Jeune from our team are here to discuss how Giskard can help ensure the quality, security, and compliance of your AI applications. We've already had an insightful morning following the round table about 'Large Language Models for businesses across customer support'. Our CTO, Matteo Dora, has discussed with Hugues Even (CDO, BNP), Guillaume Bour (Head of Enterprise EMEA, Mistral), and Anne-Laure Giret about balancing innovation in AI applications in financial services with rigorous testing and monitoring. Thanks to Artefact for organizing this event! And special thank you to HananIA Mikael OUAZAN for moderating the round table 🙌 Learn more about securing AI-based apps 🔗 https://lnkd.in/espQZTY5 #AIforFinance #AISecurity #LLMs #GenAI

🚀 Giskard will be at the AI for Finance by Artefact tomorrow! 🚀 Join our CTO Matteo Dora at 9:30 for a round table on 'Large Language Models for business across customer support' with Hugues Even (CDO at BNP Paribas) and Guillaume Bour (Head of Enterprise EMEA at Mistral AI). You can also visit us at the Amazon Web Services (AWS) booth in the GenAI zone. Guillaume Sibout from our team will be there to discuss your use cases and how Giskard can help your business ensure the quality, security, and compliance of your AI models. Book your pass here 🎟️ https://lnkd.in/eMGsp9FW 📍 Palais Brongniart, Paris ⏰ 9:30 am CET #AIforFinance #AISecurity #LLMs #GenAI