AI Safety stories
Despite high confidence in their controls, most firms still let AI agents run with broad access, and 29% have seen real harm.
With women still only 17% of cyber workers in the UK and Australia, executives warn AI could deepen security blind spots unless talent widens.
The new tools aim to curb prompt injection, data leakage and rogue AI activity as firms deploy agents with wider system access.
Enterprise security teams could see faster breach detection and response as CrowdStrike's new AI models are trained on live threat data.
Enterprises are demanding audit trails and spending controls as AI agents move from pilots into live systems with access to core business data.
Faster agentic AI adoption is forcing tighter oversight, with new controls on permissions, monitoring and cyber-risk systems.
Boards and regulators are being urged to prove critical services can hold up as frontier AI speeds attacks and shortens response times.
New controls and tests are being added as agentic AI systems spread, with Microsoft warning that risks now span tools, data and users.
Many firms lack the capacity to fix existing flaws fast enough, leaving them exposed as AI-generated attacks scale up, experts warn.
ReliaQuest says a single operator may have used a commercial AI coding agent to build malware, a console and update pipeline across 324 hosts.
Enterprises running AI agents in private clouds are getting stricter controls as Broadcom seeks to curb data leakage, prompt injection and compliance risk.
Women's skills in AI and machine learning are helping expose a cyber workforce gap, as firms race to bolster defences against smarter attacks.
Only 3% of MSPs said every client tenant is Copilot-ready, despite AI governance becoming the fastest-growing request.
Australian enterprises are under pressure to tame AI-driven complexity, compliance risk and rising technology costs as Datadog broadens its platform.
Free prompts to curb AI over-reliance are being rolled out in Brazil as PUCPR tries to preserve source citation and human judgement.
The hire aims to help Aeris capitalise on demand for secure links in autonomous vehicles, robotics and other connected machines.
Enterprise security teams are treating women's AI scepticism as foresight, as agentic systems expose the governance risks many had flagged earlier.
Victims are being left to fend for themselves as fraud becomes industrialised, with many still blaming themselves after scams.
Most financial firms using AI still lack proper risk checks, prompting fresh guidance as regulators warn oversight is lagging.
Enterprises can now keep AI workloads and sensitive data on the same private cloud, as Broadcom adds governance and security controls.
