Faye Waters

Anthropic will commit more than $100bn to AWS over the next decade, securing up to 5GW of compute to train and serve Claude, including new Trainium capacity.

IBM has launched Bob, an AI development platform covering planning, coding, testing, deployment and modernisation, with governance and audit controls built in.

Microsoft and OpenAI have amended their partnership: OpenAI can now serve its products on any cloud, while Microsoft keeps a non-exclusive IP licence to 2032.

Security updates include Red Agent, AI-BOM, and deeper integrations with Databricks, Salesforce, Cloudflare, and Google Cloud.

OpenAI’s ChatGPT Images 2.0 improves dense text and layout rendering, adds non-Latin language gains and up to 2K output via gpt-image-2 in the API.

NVIDIA and Google Cloud have extended their co-engineered AI platform at Google Cloud Next, expanding AI Hypercomputer capacity for agentic and physical AI.

Perplexity has launched the Secure Intelligence Institute, a research centre for AI security, privacy and trust led by Purdue professor Dr Ninghui Li.

Mistral AI has released Connectors in Studio, making built-in connectors and custom MCPs available via API and SDK, with direct tool calling and approval flows.

While it promises rapid vulnerability detection, the emergence of Anthropic’s latest AI raises questions about security risks, defensive preparations, and the future of knowledge work.

ArmorCode’s AI Exposure Management gives enterprises central visibility and control over AI use across applications, cloud environments and developer workflows.

Celonis will deploy its Process Intelligence platform on Oracle Cloud Infrastructure and deepen integration with Oracle Fusion applications.

IBM has launched a cybersecurity assessment for frontier model risks and IBM Autonomous Security, a multi-agent system that automates detection and response.

Wiz’s Shadow Data Detection, part of its DSPM tooling, finds stale, duplicate and redundant cloud data to cut storage costs and shrink the attack surface.

OpenAI rotated its macOS code-signing certificates and required desktop app updates after a compromised Axios package reached a GitHub Actions workflow.

Anthropic and the Australian government have signed an MOU on AI safety, covering joint evaluations, economic data and the National AI Plan.