AI Safety stories
Enterprises could cut agent coding costs and compliance risks as the new releases add server-side repository access, audit tools and spend controls.
Customers will be able to enforce zero trust controls across more AI tools as Zscaler broadens its security programme to key cloud partners.
Guardrails may not stop attackers as Anthropic's split release underscores a widening gap between AI exploit discovery and patching.
The win highlights growing demand for governed AI tools that speed up identity admin without weakening approvals, audit trails or compliance.
The deal will put Claude into banking, aviation and government systems, as DXC scales AI agents across regulated customer environments.
The proposals could shape how banks and insurers manage cyber and operational risks as AI adoption accelerates across the sector.
AI agents will be able to make purchases with user-approved controls, as Visa moves to bring tokenised payments into OpenAI's commerce tools.
Users in 10 markets can now find and connect to PureVPN servers through ChatGPT prompts, without sharing data with the chatbot.
Enterprises could gain a more standard way to compare AI risk, as the Cloud Security Alliance expands its RiskRubric ecosystem with Tumeryk.
The tie-up aims to help regulated firms move generative AI from pilots into production, while training 50,000 TCS staff on Claude.
Large companies may gain a way to move AI pilots into production, as the platform adds governance and audit controls for enterprise workflows.
Voluntary model reviews may leave gaps as advanced AI systems move closer to critical infrastructure and enterprise data.
Most UK public sector IT teams lack the infrastructure and trust needed to scale AI safely, a SolarWinds survey found.
NHS patients could be routed faster and more accurately after a UK-built model outperformed GPs and rival AI in triage tests.
The rollout gives enterprise IT teams autonomous task execution across service, security and endpoint management, with built-in privacy controls.
Rising complaint volumes and more complex cases are pushing The Ombuds Group to use AI with human oversight across all its schemes.
The appointment adds Whitehall credibility as Electric Twin pushes its synthetic audience tool into sensitive public and commercial decision-making.
Enterprise software teams are far more willing to use AI before production, with trust dropping from 82% at build to 58% at release.
The deal secures rare long-term UK AI capacity as demand for power-hungry inference computing outstrips available data centre infrastructure.
The real payoff will come from governed workflows, as executives move beyond pilots and turn AI into a measurable business capability.
