AI Safety stories - Page 2

Ledger names Ian Rogers as Chief Human Agency Officer, putting hardware approval and human oversight at the centre of its AI security push.

KnowBe4 unveils Agent Risk Manager to monitor autonomous AI agents in real time, flag prompt injections and curb rogue data access.

Infobip marks 20 years with AgentOS launch as it pushes into agentic AI, aiming to unify customer data and let AI agents act across channels.

Logicalis survey finds 77% of firms hit by cyber incidents as CIOs warn AI is adding fresh risk, with weak governance and skills shortages deepening exposure.

Forrester warns Anthropic's Project Glasswing could overwhelm vulnerability management, as AI uncovers flaws faster than patching teams can respond.

Commvault unveils AI tools to help enterprises govern agents, prepare approved datasets and recover full-stack systems after automated changes.

Doppel secures three ISO certifications for AI governance, security and privacy, as enterprise buyers demand stronger assurance against AI-driven cyber threats.

Anthropic's public beta gives developers cloud-hosted AI agents with built-in execution, tracing and governance, as rivals race to simplify deployment.

Salt warns AI agents are widening the API security gap, with 92% of organisations still short of advanced defences and 47% delaying releases.

Cloudsmith survey finds most engineering teams still lack automated SBOM checks, leaving many unready for fast EU Cyber Resilience Act audits.

Oracle unveils 25 AI-powered apps across finance, HR and customer service, as it pushes agentic software deeper into core business workflows.

Anthropic enlists Amazon, Apple and Microsoft in Project Glasswing to use Claude Mythos Preview for hunting vulnerabilities in critical software.

Only 14% of Global 2000 firms have a clear AI strategy, with research warning many are deploying tools faster than accountability and workforce training.

Logicalis study finds 89% of Chief Information Officers are learning AI as they go, with skills gaps, data issues and compliance concerns slowing roll-outs.

Nasuni pivots from cloud file storage with Active Everywhere and AI Activate, aiming to give enterprises governed access to unstructured data for AI.

Anthropic and partners unveil Project Glasswing, using Claude Mythos Preview to hunt software flaws faster amid fears of AI-boosted cyber attacks.

Anthropic’s Claude Code is under scrutiny after researchers found deny rules can weaken in long workflows, raising fresh concerns for AI-driven development.

Iran's Revolutionary Guard has named OpenAI's Abu Dhabi Stargate campus as a possible target, sharpening fears over Gulf AI infrastructure amid rising tensions.

OpenAI urges governments to prepare for AI-driven disruption as Sam Altman says rapid progress could transform work, tax and cyber security within years.

Anthropic expands a guarded AI pilot with Amazon, Apple, Microsoft and others, offering Claude Mythos Preview to hunt flaws in critical code and open source.