AI Safety stories
Despite recession fears, 74 per cent of senior executives still plan to keep AI near the top of budgets, KPMG found.
Sensitive prompts and documents will stay out of model training as ExpressVPN enters AI software with an enclave-based service for Pro subscribers.
Enterprises deploying agentic AI are getting a new tool to spot data leaks, policy breaches and runaway costs before they spread.
Teams can now block toxic or sensitive AI output before it reaches customer data, inboxes and other business systems.
Online advertising faces pressure as autonomous AI agents could soon handle searches, payments and negotiations on users’ behalf.
Greater scrutiny of generative AI is set to push observability spending up as companies seek to prove outputs are accurate and traceable.
AMD says local AI agents will need always-on PCs with more memory and compute, shifting work from apps to autonomous tasks.
Battleground-state voters overwhelmingly support tighter insurance rules and AI guardrails, citing unaffordable care and risks for children.
Most boards are using AI for routine tasks, but only 3% have woven it into risk oversight, leaving organisations exposed to fresh hazards.
Boards across software are seeking directors with AI and governance expertise as New Relic adds Wendi Sturgis to oversee its next phase of growth.
As AI moves into production, enterprises face gaps between data governance and runtime controls that can expose sensitive information and policy breaches.
Without proper oversight, rapidly growing AI agent workforces could leave firms blind to who can access systems, data and privileges.
Researchers can now report AI misuse and harmful agent behaviour under a separate programme that could expose risks in ChatGPT Agent and Browser.
Enterprises racing to deploy AI tools are risking sensitive data leaks unless security moves from discovery to runtime protection, F5 and Forcepoint say.
Pressure to adopt AI is outpacing safeguards, with most firms saying governance and legal controls have lagged behind deployment.
Longer AI-generated tracks are now available to businesses, developers and subscribers as Google widens access to its music model across products.
Users can now query AI without prompts or files being exposed, as ExpressVPN moves beyond virtual private networks into confidential computing.
The funding will help Qodo expand globally as enterprises look for ways to verify AI-written code before it reaches production systems.
The move gives Toronto AI startups access to senior academic and industry advice as they push research ideas towards commercial products.
A shortage of approved classroom AI tools is leaving most Australian teachers eager for training but unable to use them with students.
