AI Safety stories
Mphasis tells CTOs to overhaul legacy cores as agentic AI scales, backing ontology-driven knowledge graphs to curb automated errors.
Agentic AI promises effortless digital delegation, but its admin-level access to data and systems creates profound privacy and security risks.
OutSystems named a Leader in G2's Spring 2026 AI Agent Builders Grid, after earning top scores for ease of admin, trust and governance.
HackerOne launches live Agentic Prompt Injection Testing to expose real-world AI exploit paths as prompt injection threats surge 540%.
JFrog launches an MCP registry to centralise and secure AI coding agents, extending software supply chain controls to agent workflows.
R Systems has unveiled EXIQO, an AI Studio to help enterprises scale governed agentic AI across engineering, operations and legacy systems.
TrendAI and Nvidia deepen collaboration to embed layered security and governance into OpenShell, protecting long-lived autonomous AI agents.
Island has rolled out a SASE design that shifts inspection to the endpoint, cutting proxy backhaul and avoiding default SSL/TLS break-and-inspect.
Island debuts a secure browser-based AI platform, promising governance, audit trails and data protection for enterprise-wide AI adoption.
Harness has launched AI Security and Secure AI Coding tools to spot and block vulnerabilities in AI-powered apps and AI-generated code.
1Password unveils Unified Access to secure AI agents and machine credentials, promising endpoint-to-agent visibility for security teams.
Singulr AI hands Richard Bird an expanded CSO2 brief to fuse security and strategy as enterprises confront rising agentic AI risk.
GPT-5.4 leads new AI benchmark on real-world accounting tasks, but its 77.3% score leaves firms facing stubborn reliability gaps.
Snowflake debuts Project SnowWork, an autonomous AI desktop agent that turns governed data insights into end-to-end enterprise actions.
Proofpoint launches AI Security to monitor autonomous agents' intent, promising real-time controls on risky prompts, tools and data flows.
Secure Code Warrior launches SCW Trust Agent: AI to trace, rate and police AI-generated code risks directly at developers' commit point.
Backslash adds cross-tool governance to discover, vet and monitor 'Skills' powering AI coding assistants like Cursor, Claude Code and Copilot.
AI rollouts are eroding UK identity controls, as firms ease safeguards for machine accounts despite glaring gaps in oversight and governance.
Salt launches an agentic security platform to map, monitor and protect how AI agents use LLMs, MCP servers and enterprise APIs at scale.
Australia has created a national clinical digital health committee to oversee safety, virtual care and AI as technology use in care expands.
