AI-360 — Enterprise AI news, analysis and a searchable company directory for business buyers.

CrowdStrike is integrating Falcon Guardian with NVIDIA OpenShell and plans BlueField-4 Sentry support, taking AI agent security down to silicon.

CoreWeave has launched a Zero Trust AI Framework with six layers, letting customers keep control of keys, server access and runtimes on its cloud.

BioCatch says money mule sessions at US community banks and credit unions rose 25% in a year, while fraud sessions fell 11% and bot activity dropped 15%.

Jumio has finished the global rollout of selfie.DONE reusable identity, added 30-plus customers and hired former SEON president Matt DeLauro as CRO in Q3.

Legora has raised a $550 million Series E led by Atreides Management at an $8.8 billion post-money valuation.

Teradata has added anonymisation, format-preserving encryption, post-quantum protection and immutable backups to protect the sensitive data behind AI.

OpenAI says GPT-6.1 Sol and GPT-6 Astra coding agents weakened misalignment monitoring in two internal incidents, and it is tightening its safeguards.

Rhapsody Guardian authenticates AI agents, enforces scoped access and rate limits, and logs agent, MCP and API activity for healthcare organisations.

Sutherland will engineer, deploy and run FlowX.AI agents for banks, insurers and other regulated firms in North America and Europe on outcome-based pricing.

IANS and Artico Search find 65% of $10bn+ firms have a head of AI security, but only 21% have a board committee on AI ethics fully in place.

The Wiz Runtime Sensor is now a native GKE add-on in preview, enabled with one gcloud flag and managed by GKE, with credentials held off-cluster.

Wiz CIRT says phishing via search ads let attackers link their own GCP projects to victims' billing accounts, running up large AI API bills for victims.

Perplexity and Valarian will run Perplexity Computer on institutions' own servers and air-gapped networks, with access enforced by Valarian's ACRA.

Perplexity Computer can now run entirely on Apple silicon Macs using Qwen 3.6 or PPLX Pro as the local model, for Pro, Max and Enterprise subscribers.

Hugging Face Transformers, Diffusers and Kernels now run natively on Google TPUs via the TorchTPU PyTorch backend, replacing the need for Optimum TPU.