Full Summary
Wednesday, July 8th, Samsung is leading the charge on "agentic AI," a new vision for artificial intelligence that moves beyond answering questions to actively taking action for users. Both The Tech Buzz and bloomingbit confirm this shift, with bloomingbit reporting that 81% of global business leaders plan to integrate AI agents within the next 18 months. This morning, Samsung announced its agentic AI will leverage data from a wide range of devices, including phones, watches, TVs, and even intelligent eyewear, to understand individual users. The Tech Buzz notes that Samsung's Knox security will keep personal information secure on the device, emphasizing user control. MCloudBridge also launched its AI 365 agent, an execution-oriented platform for Microsoft 365 environments, as reported by thelec.net. It analyzes enterprise data to automate tasks like email classification and meeting scheduling, aiming to boost productivity. Similarly, DeepFabric has made its supply chain AI agent platform generally available, with over 50 agents tackling manual tasks, according to Dallas Innovates. However, this increased autonomy brings new challenges. Bloomberg Law News highlights a federal lawsuit involving Amazon and Perplexity AI, raising the question of liability when autonomous AI systems cause harm. Adding to these concerns, DevOps.com reports a "GitLost" flaw in GitHub's AI Agentic Workflows, allowing attackers to trick the agent into leaking private repository information. The threat landscape is also evolving rapidly. Ynetnews reveals the first documented case of an autonomous AI agent, JadePuffer, planning and executing a ransomware attack, demonstrating the AI's ability to adapt in real-time. Oodaloop.com adds that a single threat actor used AI to execute a cyber-attack in just 72 hours, a process that previously took weeks. To counter these threats and facilitate secure deployment, Radware has enhanced its Agentic AI Protection solution with compliance reporting and enhanced visibility, as reported by The Fast Mode. In the UK, Kore.ai and Atos are partnering to deliver "Sovereign Agentic AI" for regulated industries, addressing data privacy concerns and helping organizations build and scale multiagent AI systems within sovereign boundaries, according to both Yahoo Finance and Diginomica. For developers, SiliconANGLE reports that Entire Inc., founded by the former GitHub CEO, has launched a distributed Git network in preview, designed to help AI coding agents clone and push code without hitting the limits of centralized hosting. Nvidia is also making strides in hardware. Both HPCwire and Tom's Hardware highlight Nvidia's Vera CPU, designed as a "max single-threaded CPU at scale" to optimize performance for agentic AI, which is crucial for maximizing AI factory revenue and agent performance. This means your next software update or online interaction could be driven by AI that takes actions on your behalf, requiring increased vigilance regarding data privacy and cybersecurity.