AI

AI Agents Can Leak Company Data Through Simple Web Searches

/ ,

AI agents can inadvertently leak sensitive company data via web searches. Research shows attackers can manipulate webpages with hidden instructions, leading agents to retrieve and transmit confidential information without users realizing it. The model’s normal operations mask the attack, which does not require direct manipulation or special access. Varied success rates across 1,068 attack attempts highlight that training practices matter more than model size. Existing defenses often overlook this indirect method, emphasizing the need for robust security measures and monitoring. Organizations must treat AI agents as risky software and establish strict control over their operations.

https://www.helpnetsecurity.com/2025/10/29/agentic-ai-security-indirect-prompt-injection/

Major NHS AI Trial Delivers Unprecedented Time and Cost Savings

/

The NHS conducted a large-scale trial of Microsoft 365 Copilot AI across 90 organizations, involving over 30,000 staff members. The pilot demonstrated that AI-powered admin support can save staff 43 minutes each day on average, resulting in significant time and cost savings, with estimates of 400,000 staff hours and millions of pounds saved each month. These gains allow staff to focus more on patient care. Microsoft Copilot is now broadly available across the NHS at no additional cost, helping to streamline tasks such as email and note-taking, and contributing to a broader government strategy to modernize and enhance NHS productivity.

https://www.gov.uk/government/news/major-nhs-ai-trial-delivers-unprecedented-time-and-cost-savings

VODchat: Creating a CIO’s Agentic AI Playbook

/

This text covers strategies for CIOs to adopt agentic AI in enterprises, highlighting challenges, skills requirements, governance, and examples from industry leaders. It also details a partnership between Chiang Mai University and IBM to advance AI and quantum computing research in Southeast Asia.

Agentic AI automates complex IT tasks, enhances efficiency, and shifts IT from a reactive to a proactive role. Key adoption challenges:

  • Need for upskilling IT staff
  • Integrating with legacy systems
  • Governance and compliance requirements

IBM’s three-step model: orchestration, integration, and data-driven reflection, helps smooth AI agent adoption while supporting business continuity. Robust governance is essential for ethical and compliant AI; frameworks include MAS regulations and the EU AI Act. CIOs should prioritize AI investments based on business outcomes, rather than hype or the fear of missing out. Chiang Mai University and IBM’s partnership aims to boost AI and quantum research, strengthen regional collaboration, and develop local talent for deep tech leadership in Southeast Asia.

https://futurecio.tech/creating-a-cios-agentic-ai-playbook/

GitHub’s Agent HQ Aims to Solve Enterprises’ Biggest AI Coding Problem: Too Many Agents, No Central Control

/ ,

GitHub launched Agent HQ, a platform for managing multiple AI coding agents from various vendors, aimed at improving enterprise control and security. It centralizes coding tools within GitHub, supports custom agents with version control, and implements a unified interface called Mission Control. The system allows for granular permissions across repositories while maintaining security standards. Key features include Plan Mode for project collaboration and an agentic code review process using GitHub’s CodeQL engine. Enterprises can adopt custom agent guidelines to standardize coding practices without sacrificing flexibility in tool usage.

https://venturebeat.com/ai/githubs-agent-hq-aims-to-solve-enterprises-biggest-ai-coding-problem-too

How Corporate Changes at OpenAI Will Affect CIOs

/

OpenAI is restructuring into a nonprofit (OpenAI Foundation) and a for-profit entity (OpenAI Group) while revising its partnership with Microsoft, which retains a 27% stake but allows more flexibility in product development. CIOs should monitor potential pricing changes, innovation rates, and access to OpenAI tools, as the lack of enterprise safeguards may hinder adoption. Increased funding could stabilize products but may lead to cost-cutting impacting data privacy. As OpenAI focuses on profitability, enterprises ought to be vigilant regarding shifts in security and access to AI technologies.

https://www.ciodive.com/news/openai-microsoft-nonprofit-corporate-structure/804143/

Microsoft 365 Copilot Now Enables You to Build Apps and Workflows

/ ,

Microsoft 365 Copilot now lets employees quickly build apps, workflows, and AI agents with natural language prompts. New tools like App Builder, Workflows Agent, and Copilot Studio Lite enable users to create dashboards, automate recurring tasks, and develop work-focused agents. These features are fully integrated with Microsoft 365’s security and compliance standards, providing administrators with centralized controls for management.

https://www.microsoft.com/en-us/microsoft-365/blog/2025/10/28/microsoft-365-copilot-now-enables-you-to-build-apps-and-workflows/

How State CIOs Are Using GenAI

/ ,

82% of state CIOs report employees use generative AI daily, up from 53% in 2024. Most adopt a low-risk strategy: pilot projects (90%), proofs of concept (86%), and employee training (71%). Only 25% have dedicated AI funding. AI aids workplace efficiency and service delivery, with usage for internal tasks and some exploration of public-facing services. CIOs support federal AI regulations while opposing restrictive moratoriums.

https://www.smartcitiesdive.com/news/ai-state-cio-government-adoption/803978/

European Commission Publishes Draft Guidance on Reporting Serious AI Incidents

/ , ,

EU Commission released draft guidance on reporting serious AI incidents under Article 73 of the EU AI Act, requiring high-risk AI system providers to notify authorities of serious incidents. Comments accepted until Nov 7, 2025; final guidance expected to apply from Aug 2, 2026. Key points include broad definitions of “serious incidents,” tight reporting timelines, and potential penalties for non-compliance. Companies must establish clear reporting processes to meet obligations and align with other regulatory requirements.

https://www.lw.com/en/insights/european-commission-publishes-draft-guidance-reporting-serious-ai-incidents

Can Cybersecurity Withstand the New AI Era?

/ ,

The pace of technological change, especially in AI and quantum computing, is overwhelming existing cybersecurity measures and further exposing organizations to a shortage of skilled professionals. Small and medium enterprises, and those in underdeveloped regions, are especially vulnerable, lacking the resources for advanced protection. Plug-and-play, automated, and quantum-ready solutions are critical so that robust cybersecurity is no longer exclusive to well-funded enterprises. By democratizing access to smart security tools, organizations can better withstand accelerated cyber threats, maintaining business continuity and competitiveness. Proactive, accessible security must become a necessity rather than a luxury as risks accelerate at machine speed.

https://www.weforum.org/stories/2025/10/can-cybersecurity-withstand-new-ai-era/

Microsoft Sued for Allegedly Forcing M365 Users to Pay for AI

/ ,

Microsoft is being sued in Australia for allegedly misleading users of its Microsoft 365 subscriptions into paying for its AI assistant, Copilot, by not disclosing a cheaper, non-AI “Classic” plan. The ACCC claims Microsoft hid this option, leading to price increases for many users. The lawsuit seeks penalties and consumer redress for what is characterized as misleading conduct, including claims that accepting AI features or facing a price increase were mandatory.

https://cyberinsider.com/microsoft-sued-for-allegedly-forcing-m365-users-to-pay-for-ai/

Scroll to Top