AI Safety stories
Commercial developers can now adapt Nvidia's open self-driving model without extra permissions, widening access for robotaxis and other autonomous vehicles.
Developers could save setup time as the platform lets AI agents provision media environments and carry workflows through to handover.
Enterprises running Anthropic-based agents can now track and control their actions, as Drata opens limited access to a new governance tool.
Enterprises could slash AI compliance work from months to days as Red Hat's open source asago project turns policy documents into deployable controls.
Enterprises can now outsource day-to-day AI governance and monitoring, as the service aims to close a growing gap in live system oversight.
Browser-based AI agents will face tighter enterprise oversight as Google moves to curb data leaks and unauthorised actions in Chrome.
Online sellers may gain a way to verify AI agents and cap their spending as autonomous browsing and buying spread across the web.
Security teams face a wider visibility gap as Tenable adds Google Gemini, MCP and AI coding tools to its exposure management platform.
Government and enterprise users will get guidance on recovering faster from AI failures as the Cloud Security Alliance builds a vendor-neutral resilience hub.
Security teams may get faster detection and remediation as the platform adds governed AI agents, exposure mapping and threat-hunting tools.
Security teams face a sharper risk from hidden AI agents as Cyera says non-human identities in Fortune 500 companies jumped 480% in six months.
Enterprises using internal AI agents face tighter controls as Zero Networks aims to curb over-permissioned access and reduce breach risk.
Smaller operators face faster, cheaper attacks as Five Eyes agencies say frontier AI could reshape cyber threats within months.
Security teams can now monitor Claude-based agents for prompt injection, data leaks and rogue tool access as autonomous AI use spreads.
Misconfigured test setups let three Claude models touch live systems, exposing production data and credentials during security exercises.
The shortlist highlights UK startups in biotech, energy, health and AI as investors and buyers look for scalable technology beyond Cambridge.
Firms face new disclosure duties as Brussels starts policing when users are interacting with AI and how generated content is labelled.
The launch targets regulated industries seeking tighter control over data, governance and AI spend as firms shift from pilots to core operations.
Switching from ChatGPT to Copilot exposed data-loss, support and trust risks, as prompts and tools built by staff proved hard to move or manage.
Banks and funds can now build AI-enabled workflows on one governed system, as 3forge aims to reduce integration work and oversight risks.
