The tech industry is cautiously optimistic about Gold Eagle, a US government initiative to manage the surge of AI-discovered software vulnerabilities. The project aims to create a centralized clearinghouse for these vulnerabilities, but experts worry it could overwhelm IT teams if not executed well. Testing with Anthropic's Mythos LLM has already uncovered 10,000 significant vulnerabilities in a single month, highlighting the scale of the challenge.

Microsoft has announced a partnership with AMD to deploy the latter's Helios rack-scale AI accelerator in its data centers, including Azure. The accelerator, which uses Instinct MI455X GPUs and 31.1 TB of HBM4 memory, is expected to compete with Nvidia's Vera Rubin NVL72 system. Microsoft will also add a new VM series using AMD's sixth-generation Epyc CPUs and integrate Pensando DPUs into Azure Boost.

Many companies have integrated AI into workflows to boost productivity, but often fail to leverage AI's full potential in transforming enterprise data ecosystems, writes Jiayan Yin, a data engineer at Tiktok. Data agents, unlike chatbots, can perform complex tasks such as generating SQL queries and interpreting data, but relying solely on them can lead to issues such as ambiguous terminology and inconsistent answers.

AI is becoming integral to modern supply chains, with businesses in retail and manufacturing deploying AI, generative AI and agentic AI to automate workflows, manage contracts and optimize inventory. While autonomous planning remains a goal, most organizations currently benefit from foundational AI applications that improve forecasting, exception management and workflow automation.

Capital One has released VulnHunter, an AI-powered security tool, as open source to help find and fix software vulnerabilities at the code level. Unlike traditional vulnerability scanners, VulnHunter uses agentic reasoning to identify exploitable defects and propose targeted code remediation.

US companies face challenges in hiring for AI fluency, with 59% reporting a bad AI hire in the past year despite 95% of companies making AI fluency a formal requirement. Many companies set the fluency bar at basic tool awareness, leading to frequent AI-driven errors. Organizations can improve by redefining AI fluency, using a shared rubric for assessment and piloting structured screening processes.

Chief medical officers across various healthcare organizations anticipate significant changes for physicians over the next two years, driven primarily by the integration of AI. AI is expected to ease administrative burdens through ambient documentation and decision support, allowing physicians to focus more on patient care. However, challenges such as increased documentation demands, administrative burdens and workforce constraints remain. The evolving role of physicians will also include leading AI-enabled clinical ecosystems and team-based care models, underscoring the need for strategic use of technology to improve efficiency and patient outcomes.