AI Safety stories

Harvard research finds “human-like” AI mirrors Western liberal values, raising concerns over cultural bias as it spreads worldwide.

OpenAI launches GPT-5.4 with “Thinking” mode and native computer control, targeting advanced coding, office workflows and safer automation.

Forcepoint adds ARIA AI assistant and a faster endpoint agent to Data Security Cloud to tighten policy control for generative AI workloads.

In 2026, AI agents move from pilots to the enterprise core, forcing firms to prove measurable value under tight governance.

RecordPoint launches an MCP Server to give AI tools secure, auditable access to governed enterprise data via a standardised interface.

ExpressVPN launches a beta MCP server that lets AI tools monitor VPN status and switch regions directly from its desktop apps.

CIOs warn AI is being rolled out faster than governance can keep up, with many fearing security gaps and lacking oversight of tools in use.

Women shape our world, yet lack power over AI systems that govern work, wealth and welfare - where are the true godmothers of AI?.

Agentic AI massively accelerates elite cyber teams but can slow inexperienced hackers, Hack The Box's large-scale benchmark reveals.

Women in cybersecurity, long trained to question and validate, are uniquely placed to lead the era of risky, fast‑moving AI tools.

RecordPoint launches an MCP Server to give AI agents standard, auditable access to governed enterprise data without bespoke connectors.

Agentic AI promises autonomy but raises urgent questions over whose values guide its decisions and how equity is safeguarded in practice.

Check Point launches Secure AI Advisory to tie AI governance and regulatory readiness directly into enterprise security operations.

On International Women's Day, CSA spotlights women steering its AI Safety Initiative, proving inclusion is core infrastructure for secure AI.

CIOs say AI adoption is racing ahead of governance, with skills gaps, risk fears and sustainability concerns stalling efforts to scale.

Microsoft's new Cyber Pulse briefing warns ungoverned AI agents and 'double agents' pose rising security and compliance risks for firms.

Manifest research reveals executives overestimate AI security readiness, as AppSec teams warn of unmanaged tools, blind spots and rising risk.

Celonis says its process intelligence platform has helped customers save USD $10 billion by underpinning AI-driven operations and automation.

OpenAI unveils GPT-5.3 Instant, promising smoother, more direct ChatGPT conversations, fewer refusals and improved accuracy on web-sourced facts.

Non-consensual deepfake abuse of women is not just online harassment but cybercrime, demanding full threat intelligence and security action.