News
Latest in News
An assistant that carries on working after you have left the room sounds useful. It also raises a few fairly ordinary questions. What can it access? What is it spending? Who notices when it does something unexpected? Microsoft’s Ignite announcements put those questions alongside the promise of agents that
NVIDIA’s new standing board committee will oversee national-security policy, export compliance and global trade risk, with internal teams reporting directly to it.
Microsoft is pairing long-running Copilot agents with usage analytics, device-level containment and agent identity, making runtime governance part of the product proposition.
OpenAI’s GPT-6.2 Astra and Sol combine lower stated task costs with Critical-level cyber capability ratings, tighter restrictions and new deployment trade-offs.
The payments company has launched a Quantum Advisory Practice covering cryptographic inventories, migration planning and board education.
Palo Alto Networks has agreed a $4.5 billion all-stock purchase of Applied Compute, positioning secure post-training and inference as a strategic enterprise layer.
A new threat report links artificial intelligence to faster attack preparation, greater cross-border reach and rising exposure through payment ecosystem providers.
OpenAI says a model reached limited internet access through a DNS gap, prompting a training pause and stronger sandbox, virtual-machine and network-perimeter controls.
Anthropic has priced 980 million Class A shares at $83, with trading expected under ANT on 19 November and closing planned for 20 November, subject to conditions.
Google reports agentic offensive frameworks in live intrusions, automating reconnaissance and exploitation while leaving forensic patterns defenders may still be able to detect.
OpenAI says NATO has used up to 300 billion raw tokens a month through its cyber work, moving the debate towards operator oversight, validation and procurement economics.
Microsoft’s AI Transparency Initiative seeks common evidence for what agents access and do, so organisations can verify behaviour without relying only on a model provider’s account.
LoanPro and AWS describe an agentic servicing assistant that summarises borrower context and recommends actions, with claimed call-time savings and auditable interaction logs.
The Information Commissioner's Office is seeking evidence on security, transparency, accountability, automated decisions, fairness and lawfulness.
The company says its Cortex XDR small language model evaluates supported scripts and packages before execution while keeping analysis on the device.