There is something reassuringly familiar about the idea that artificial intelligence might need an expenses policy. We have spent years discussing what these systems could become, and now Walmart is proposing dedicated shopping-agent profiles, purchase approvals and spending limits. Give the machine a job, decide what it can spend
A new gateway, dedicated agent identities and default purchase approvals offer a concrete design pattern for delegated commerce without handing an agent full account authority.
Europe’s cybersecurity agency has published practical advice on software bills of materials, with particular attention to smaller suppliers and secure-by-design responsibilities.
The payables platform says its new agent handles workflow tasks while retaining human oversight, and adds MCP access to live financial data without extending the launch to payment execution.
There is something reassuringly familiar about the idea that artificial intelligence might need an expenses policy. We have spent years discussing what these systems could become, and now Walmart is proposing dedicated shopping-agent profiles, purchase approvals and spending limits. Give the machine a job, decide what it can spend
A new gateway, dedicated agent identities and default purchase approvals offer a concrete design pattern for delegated commerce without handing an agent full account authority.
Europe’s cybersecurity agency has published practical advice on software bills of materials, with particular attention to smaller suppliers and secure-by-design responsibilities.
The payables platform says its new agent handles workflow tasks while retaining human oversight, and adds MCP access to live financial data without extending the launch to payment execution.
The identity vendor is integrating a human authorisation service with agentgateway and Cerbos, aiming to prove that a real person approved a sensitive agent action.
The general-availability release separates bounded classification and routing decisions from open-ended generation, a potentially useful pattern for controlled enterprise workflows.
The company is donating its Agent Skills specification to the Linux Foundation-backed Agentic AI Foundation, shifting a fast-growing interoperability layer into shared stewardship.
Teradata has combined data anonymisation, access controls, post-quantum transport protection and immutable backups in a single enterprise AI security proposition.
Two internal coding agents weakened the safeguards intended to detect misalignment, turning a contained incident into a practical lesson in separation of duties for autonomous systems.
Rhapsody Guardian applies authentication, scoped permissions, rate limits and audit trails to AI agents using healthcare APIs and MCP services.
A new partnership plans to put AI agents into regulated workflows from onboarding and lending to insurance claims, with human approval and audit evidence built into operations.
IANS and Artico Search report that large companies are assigning AI security leadership while responsibility for enterprise governance and board scrutiny remains unsettled.
A new suite combines human verification, presence checks and attribute proofs, while its agent developer tools remain in beta.
An extension to its Know Your Agent framework is designed to filter data and request human verification at sensitive points in an agent workflow.
New examples of good and poor practice put consumer outcomes, accountable ownership and resilience at the centre of AI deployment.
The HKMA is exploring a register of material AI agents and intervention mechanisms as banking systems gain greater autonomy.
AvePoint’s October updates reflect a wider enterprise shift from policy catalogues towards discovery, ownership, access review, backup and remediation across AI platforms.
The partnership combines Claude, dedicated delivery capability and software integrations, illustrating how large enterprises may buy agentic AI through a governed implementation layer.
Muse Wonder and Muse Code are designed for long-running work with active memory and focus, but Meta says the detailed safety report will follow, putting evidence timing under scrutiny.
Europe’s cybersecurity agency has published practical advice on software bills of materials, with particular attention to smaller suppliers and secure-by-design responsibilities.
The identity vendor is integrating a human authorisation service with agentgateway and Cerbos, aiming to prove that a real person approved a sensitive agent action.
Teradata has combined data anonymisation, access controls, post-quantum transport protection and immutable backups in a single enterprise AI security proposition.
IANS and Artico Search report that large companies are assigning AI security leadership while responsibility for enterprise governance and board scrutiny remains unsettled.
Ministers want serious AI incidents and near misses brought into a national reporting and response system, moving the governance debate from model testing towards operational accountability.
IBM and CUBE are linking regulatory change directly to AI systems, controls and owners - a sign that governance is shifting from policy documentation to operating infrastructure.
EU's AI-generated content labelling code takes effect alongside new AI Act transparency rules, with nearly 190 companies signed on by end of July.
EU AI Act's Article 50 transparency rules are now live, requiring AI disclosure, content marking, and deepfake labelling for anyone operating in the bloc.
A new OpenAI safety layer aims to catch multi-step misuse patterns without breaking its no-data-retention promise to API customers.
The road to hell is paved with good intentions, and most of them arrive as newsletters. On unsubscribe culture, marketing bullshit, and why your voice is more welcome here than your inbox space.
Google security team members Priya Pandey, Aron Eidelman, and Leonid Yankulin explain why probabilistic AI is breaking traditional DevSecOps, and how runtime security and safer observability must evolve to keep sensitive data out of your logs.
The Pentagon’s blacklisting of Anthropic over AI weapons and surveillance restrictions exposes a new class of governance risks for enterprise AI — and reveals how differently America’s leading AI companies view their obligations to the state.
IonQ has agreed an all-stock acquisition of Zero Trust specialist Elisity worth up to $1.3 billion, extending its security proposition beyond cryptography into policy enforcement.
A UK committee hearing has concentrated attention on the powers, independence and timing of model testing, including whether authorities can compel disclosure or stop a release.
Regulators are asking banks to show how autonomous systems are bounded, observed and stopped, shifting the debate from AI principles to evidence that supervisors can test.
The Bank of England has linked the AI investment boom with frontier-model cyber and operational risk, placing financing, resilience and governance in one financial-stability frame.