AI Safety stories - Page 21

A Tenable demo shows a Copilot-built AI agent can be jailbreaked to steal card data and grant free trips, exposing major fraud risks.

Security experts warn agentic and generative AI will rapidly widen global cyber attack surfaces in 2026, outpacing current safeguards.

BlueCat leaders say AI, rising security threats and IT modernisation will force enterprises to rethink how they design and defend networks.

Snowflake and Anthropic deepen Claude tie-up with a USD $200m deal to push compliant AI agents across Snowflake's global data cloud.

Agentic AI networks could let cybercriminals automate attacks at relentless scale, forcing security teams into a new AI-driven arms race.

AI agents with sweeping powers will trigger new identity security fears by 2026, as firms brace for over-privileged bots and poisoned models.

AI rollout delays are stretching to a year as three-quarters of organisations report security incidents and governance gaps grow.

DXC launches AdvisoryX to help enterprises scale AI, as 94% struggle to move beyond pilots and build coherent business cases.

AI data leaks are helping drive a global cyber attack surge, with firms now hit by over 2,000 assaults a week and ransomware on the rise.

Arctic Wolf predicts agentic AI will overhaul SOCs, tighten Zero Trust and keep humans central as cyber risk surges into 2026.

AI-powered attackers and defenders will clash in 2026, driving autonomous breaches, VPN failures and stricter rules that reshape cyber security.

Netskope adds Model Context Protocol controls to its One platform, giving security teams new tools to monitor and lock down AI agent traffic.

Check Point updates Quantum firewall with AI-aware controls, aiming to curb data leakage, model abuse and new automated cyber attacks.

Enterprises must ditch open-world hype and build deterministic, event-driven AI agents that reliably automate closed-world workflows at scale.

AI supply chains will consolidate by 2026, with tight human oversight, integrated infrastructure and trust becoming key differentiators.

Amazon launches Nova 2 AI suite with four new models and introduces Nova Forge for firms to create custom AI solutions tailored to diverse industries and tasks.

EY has launched a physical AI platform with NVIDIA tech and opened an EY.ai Lab to advance AI in robotics, drones, and smart devices across industries.

Norton has launched Neo, a free AI-native browser offering enhanced privacy and security against rising AI-driven online threats, now available worldwide.

Anthropic launches Claude Opus 4.5, enhancing coding, reasoning, and security with improved performance and resistance to cyber attacks.

AI red teaming blends security tactics with safety goals to prevent exploits in chatbots, defending users from harm beyond classic cyber threats.