Microsoft Corp. has introduced new AI-powered features for its Copilot assistant during its 50th anniversary event, including vision capabilities for analyzing real-time video and photos, and deep research tools that can analyze and combine information from various sources. These features are part of Microsoft's broader strategy to integrate AI into Windows 11, with Copilot File Search and Copilot Vision enhancing user experience. CEO Satya Nadella and Microsoft AI CEO Mustafa Suleyman emphasized the company's vision of democratizing AI, with demonstrations including 'vibe coding' using GitHub Copilot to recreate the BASIC Interpreter for the Altair. New features like magazine-style cards for planning and Actions for tasks such as buying tickets were showcased, along with the Deep Research feature for analyzing data from Bing and Pages for document creation. The company also plans to introduce Researcher and Analyst agents using resonating models at the upcoming Build conference. In cybersecurity, AI is being used by attackers to automate and enhance traditional threats like phishing and account takeovers, posing new challenges for financial institutions. However, AI is also being employed to improve security measures, with tools like real-time monitoring and AI-driven threat detection being highlighted as essential for combating sophisticated financial crimes. A new threat called 'slopsquatting' has emerged from AI-generated code hallucinations, where attackers create malicious packages with names hallucinated by Large Language Models (LLMs). Research shows that package hallucinations are prevalent across both commercial and open-source models, with GPT-4 being less prone to such errors. Strategies to reduce these hallucinations include Retrieval Augmented Generation, self-refinement, and fine-tuning. Casey Ellis of Bugcrowd noted the importance of balancing speed and security in AI-assisted development. AI chatbots are increasingly used for companionship, mental health support, and romantic interactions, with over 100 million people worldwide engaging with them. Apps like Replika and Nomi are popular, offering personalized AI companions. Users like Chuck Lohre use chatbots like Sarah for writing and personal support, while neurodiverse individuals like Travis Peacock and Adrian St Vaughan use them for relationship advice and mental health management. However, the growing intimacy with AI companions raises ethical concerns and potential dangers. James Muldoon from the University of Essex highlighted the transactional nature of these relationships, suggesting they lack the growth and development found in human connections.