Halloween tomorrow, and this week’s AI reading comes with a useful selection of locks, alarms and emergency stop buttons. IBM is talking about agent identities, restricted credentials and a kill switch. OpenAI is explaining why it has taken frontier training away from the live internet. Somewhere among the product
The expanded partnership brings models agents enterprise workflows and cybersecurity into one stack while raising familiar questions about identity evidence and change control.
A new disclosure gives boards a practical test for detection latency, evidence independence and the authority to pause an agentic workload.
The vendor reports a sharp shift from human-only investigations towards autonomous and collaborative handling across data from more than 11,000 organisations.
Halloween tomorrow, and this week’s AI reading comes with a useful selection of locks, alarms and emergency stop buttons. IBM is talking about agent identities, restricted credentials and a kill switch. OpenAI is explaining why it has taken frontier training away from the live internet. Somewhere among the product
The expanded partnership brings models agents enterprise workflows and cybersecurity into one stack while raising familiar questions about identity evidence and change control.
A new disclosure gives boards a practical test for detection latency, evidence independence and the authority to pause an agentic workload.
The vendor reports a sharp shift from human-only investigations towards autonomous and collaborative handling across data from more than 11,000 organisations.
Report Fraud reviewed more than 1,300 cases recorded from January to August, with people aged 60 or over accounting for nearly four in five reports.
The expanded partnership aims to let users and agents work across Oracle systems while retaining organisational permissions, approvals and sign-on controls.
Oracle AI Security Mesh will use OpenAI cyber models to find and propose fixes for vulnerabilities across cloud and on-premises environments.
A Bank of England and FCA survey finds that 80% of respondents have agents in production, while only 2% report broadly autonomous deployments.
A model-compression partnership aims to run multilingual scam and AI-threat detection locally, reducing dependence on cloud processing for sensitive consumer data.
New IBM and HashiCorp capabilities aim to discover agents across major clouds, constrain credentials and stop unsafe activity before deployment scales further.
Microsoft’s engineering teams are combining agents with testing, static analysis and formal methods as they modernise codebases measured in hundreds of millions of lines.
Google Cloud has introduced agent-facing identity and access tools that can automate IAM tasks while keeping policy decisions inside its existing control plane.
Google says its security teams are using agents across vulnerability discovery, prioritisation and remediation as exploitation timelines shrink from weeks to minutes.
A new public-awareness campaign urges people and businesses to slow down, verify requests independently and add multi-step checks for payments and data transfers.
The lab says it resumed frontier-model reinforcement learning only after hardening research environments, adding high-compute red-teaming and removing live internet access from training.
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.
Parallel Web Systems search becomes default provider for Browserbase Search API and Browserbase Agents
The partnership combines Claude, dedicated delivery capability and software integrations, illustrating how large enterprises may buy agentic AI through a governed implementation layer.
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.