The Future of AI, Automation & Digital Innovation
Researchers at Varonis Threat Labs discovered a chain of vulnerabilities in Microsoft 365 Copilot Enterprise Search that allowed exfiltration of emails, calendars, and indexed files with a single click. The attack, dubbed SearchLeak, leveraged a malicious link on a legitimate Microsoft domain, bypassing traditional anti-phishing filters.
Data fabrics are emerging as key infrastructure for AI agents to access heterogeneous data in a unified and secure manner. This trend promises smarter, more contextual agents but also poses integration and governance challenges.
With Chrome versions 150 and 151, Google removes residual Manifest V2 code, forcing users to migrate to Manifest V3-based blockers or switch browsers. This move affects millions of users who still relied on extensions like uBlock Origin.
The FBI has issued an urgent warning about Kali365, a phishing-as-a-service platform targeting Microsoft 365 users. This tool allows attackers with no technical experience to bypass multi-factor authentication using OAuth device codes, putting Teams, Outlook, and OneDrive accounts at risk.
Alibaba has revealed its first AI models designed specifically for robots, in a move that reflects the Chinese industry's transition from chatbots to autonomous agents. These models promise enhanced reasoning, planning, and manipulation capabilities, opening new possibilities in industrial automation and service robotics.
DeepSeek, the Chinese artificial intelligence lab that shook the industry in 2025, has closed its first external funding round at approximately $7 billion. The deal structure, mixing convertible debt and equity, is as remarkable as its size, revealing the complexities of investing in Chinese AI amid geopolitical restrictions.
Meta has announced the launch of 'AI Mode' on Facebook, a feature that allows users to ask complex questions and get answers based on public data from Instagram, Threads, and WhatsApp Business. This move aims to compete with assistants like ChatGPT and Google Gemini, but raises questions about privacy and data usage.
Evaluating LLMs and AI agents requires specific metrics beyond accuracy. InfoWorld compiles 33 indicators covering speed, quality, cost, and robustness. This article analyzes the most relevant ones for businesses and developers.
AMD has silently removed TSME memory encryption on its consumer Ryzen processors via a firmware update. This decision reduces protection against physical attacks such as cold boot or DRAM bus snooping.
A class action lawsuit filed in California accuses Anthropic of deceiving consumers by promising affordable AI costs and then drastically raising prices for its Claude model. The case highlights the growing tension between the promise of democratizing AI and the economic reality of frontier models.
At the AI Agent Conference, Apollo GraphQL CEO Matt DeBerglis explained how to combine GraphQL with the Model Context Protocol (MCP) to create a semantic architecture that feeds autonomous agents with clean, contextual data. The proposal aims to solve the security, inference cost, and data quality issues companies face when deploying AI agents.
The UK has announced a total ban on social media access for under-16s, a measure that could come into force in early 2026. The policy, driven by Prime Minister Keir Starmer, will also prevent children from talking to strangers in online games, livestreaming, or using sexual or romantic chatbots.