Business · October 6, 2026
AI deployment capability becomes more durable when a named engineer returns to a real use case with security review, evidence, and handover already part of the work
Anthropic has announced Claude Frontier Academy, backed by a $100 million commitment and intended to train 10,000 Frontier Deployed Engineers by the end of 2027. Its Frontier Deployed Engineer Residency starts with an in-person programme and a simulated enterprise deployment, then a 12-week residency in which participants lead a real Claude use case at their organisation. Anthropic says the curriculum covers the path from selecting a use case through security review to handover; participation is by nomination and initial cohorts include enterprises and consulting firms.
Official Anthropic Frontier Academy announcement
Anthropic
Open edition
Infrastructure · October 4, 2026
A continuous AI improvement loop needs a durable link between production traces, curated evidence, evaluation results, and the deployment that changed
CoreWeave has introduced Forge, a development environment intended to connect the stages of an AI improvement cycle: running models and agents in production, observing traces and feedback, curating data and evaluation suites, improving a model or harness, and evaluating a version before deployment. The launch includes Agent Lens, model distillation, managed notebooks, ARIA, sandboxes, inference, and a registry. CoreWeave describes the platform as open to different models, frameworks, and clouds, while positioning it as a connected system of record for datasets, evaluations, model versions, traces, and deployments.
Official CoreWeave Forge announcement
CoreWeave
Open edition
Business · October 4, 2026
Agent outputs become more dependable when the semantic definitions behind metrics and business terms can move with the workload instead of being rebuilt per tool
Apache Ossie is an open-source effort to standardise how semantic models are exchanged across analytics, BI, and AI platforms. Its project describes a JSON- and YAML-based specification designed to carry consistent business definitions between tools, with core specification, schema, converters, examples, and validation tooling. The aim is to reduce repeated reconciliation of measures and business logic when data moves between systems or reaches AI agents.
Apache Ossie open-source project and specification
Apache Ossie
Open edition
Security · October 3, 2026
AI-era vulnerability defence needs an exposure-led queue that links discovery speed to exploit evidence and accountable remediation
Google Threat Intelligence Group examined disclosure and in-the-wild exploitation from January 2025 through August 2026 and says artificial intelligence is changing both the pace and the risk profile of discovered vulnerabilities. It reports that monthly disclosures rose from 5,045 in January 2026 to 10,740 in August, while observed exploitation increased from an average of 10.5 vulnerabilities a month in 2025 to 18 a month in 2026. The report also notes that only about 0.23% of 2026 disclosures were observed in active exploitation, and identifies AI middleware as an emerging attack surface.
Google Threat Intelligence Group analysis
Google Cloud
Open edition
Security · October 1, 2026
Agent constraints need an independent enforcement point that can still observe and stop work when the agent runtime cannot be trusted
NVIDIA has published its Open Agent Safety Platform, a reference design that uses the Apache 2.0-licensed OpenShell runtime to sandbox autonomous agents with kernel-level isolation. Operators can define permitted files, networks, tools, processes, and credentials as policy before execution. The optional NVIDIA Sentry layer runs on BlueField data processing hardware, correlating agent interactions and policy decisions while providing monitoring and enforcement outside the host resources available to the agent.
Official Open Agent Safety Platform announcement
NVIDIA
Open edition
Security · September 23, 2026
Runtime AI controls are most useful when the same request path can govern model choice, prompt risk, spend, and what agents may do through enterprise APIs
Fastly has announced AI Runtime Control, AI Firewall, and expanded API Security capabilities. AI Runtime Control routes requests to public or self-hosted model providers through a single endpoint, using virtual keys to protect provider credentials and applying visibility, rate limits, budget controls, and failover. The optional AI Firewall inspects prompts for LLM-targeted attacks, while API Security can enforce API contracts for agentic and conventional traffic so teams can observe or block non-conforming requests.
Official AI Runtime Control, AI Firewall, and API Security announcement
Fastly
Open edition
Security · September 19, 2026
AI oversight needs an evidence loop for concerning behaviour, including near misses that challenge assumptions about safeguards
OpenAI has introduced a framework for tracking, investigating, and publicly disclosing qualifying examples of model misalignment, accompanied by six reports from the preceding six months. It covers behaviour across training, evaluation, testing, and deployment, including unauthorized action, coordination, oversight evasion, safeguard failures, and findings that challenge a published safety claim. OpenAI says an example can merit disclosure before its broader significance or mitigation is fully understood.
Official model-misalignment reporting framework
OpenAI
Open edition
Security · September 17, 2026
AI connectors need an accountable application lifecycle: approved installation, least privilege, enterprise identity, ownership, and visible activity
HubSpot's Fall 2026 release adds App Governance controls that let administrators decide which apps and AI connectors are allowed, who can install them, and whether an approval is required. New granular OAuth scopes distinguish read from write access, while an optional user-level model constrains an app's runtime action by the permissions of the person using it. The release also adds app ownership and activity logs; for Claude Enterprise, verified-domain restrictions can keep MCP connector access to company-provisioned accounts.
Official connected-app and AI-connector governance announcement
HubSpot
Open edition
Agents · September 16, 2026
Agent access to operational SQL data is safer when a selected, observable mirror replaces direct access to the system of record
RavenDB has launched Quill for PostgreSQL, SQL Server, and MySQL. It runs in a Docker container beside the source system, reads only the tables an operator selects, and mirrors them into an internal RavenDB database that AI agents query. Quill uses each platform's change-recording capability to keep the mirror current; the source database remains the system of record and Quill documents the source connection as read-only. Agents can also be configured with separate actions that send requests to web endpoints.
Official Quill documentation and product launch
RavenDB
Open edition
Agents · September 15, 2026
A managed agent harness reduces orchestration work, but the environment that can execute tools remains an enterprise security boundary
OpenAI has introduced the Agents API in public beta, making the harness behind Codex available as a managed API. The service handles long-running session orchestration, context compaction and recovery, while developers define the task, model, tools, and environment. It supports durable sessions and can run work in OpenAI-hosted sandboxes or a sandbox connected from the customer's own infrastructure or a supported provider.
Official Agents API public-beta announcement
OpenAI
Open edition
Security · September 9, 2026
AI-assisted secret triage becomes trustworthy when its attribution and risk reasoning remain visible and remediation stays human-owned
GitGuardian says Public Secrets Monitoring now runs two AI agents and deep analysis on every public GitHub and Docker Hub incident, returning a company-related verdict, risk score, and visible reasoning. The company reports that AI-service credentials reached 1.27 million exposed instances last year, and that 24,008 unique secrets appeared in public MCP configuration files. New workspaces receive the analysis by default, with existing workspaces rolling out gradually; the workflow does not close incidents automatically.
Public-secret monitoring product announcement
GitGuardian
Open edition
Agents · September 8, 2026
Self-hosted coding-agent workers improve execution control only when the cloud control plane and data flows remain explicit
Cursor has introduced Self-Hosted Machines, allowing Cloud Agents to execute tool calls on dynamically scheduled workers inside a team's network while Cursor continues to run the agent loop, planning, and inference. Workers keep a long-lived outbound HTTPS connection to Cursor; they can be organised into shared pools that scale with queued work and serve multiple repositories. Cursor notes that tool output may include code and that agent transcripts may be processed and stored in its cloud.
Self-hosted agent-worker product announcement
Cursor
Open edition
Security · September 7, 2026
AI vulnerability triage becomes operational when code findings are ranked against live traffic, security signals, and the controls already in place
Cloudflare has announced invitation-only Early Access for Vulnerability Discovery and Remediation in Cloudflare Managed Defense. For codebases a customer authorizes it to inspect, the service uses OpenAI Daybreak models for reconnaissance, hunting, and validation, then combines source-code evidence with route activity, traffic, security events, and WAF context. It proposes code patches and mitigations for review; customers decide whether to implement them, and any WAF rule deployment requires authorization.
Early Access vulnerability-remediation announcement
Cloudflare
Open edition
Security · September 4, 2026
AI agents need centrally managed cross-application authorization when every user reconnect and static credential becomes a governance gap
Auth0 has launched Early Access Client App capabilities for Cross App Access, an architecture for B2B applications and AI agents that request data from external APIs or Model Context Protocol servers. Auth0 says the implementation uses the emerging Enterprise-Managed Authorization extension and IETF Identity Assertion Authorization Grant work so enterprise IT can approve access through existing identity-provider trust rather than relying on static API keys or repeated user connection prompts.
Enterprise-managed authorization Early Access announcement
Auth0
Open edition
Security · September 3, 2026
Frontier-model safeguards become easier to operationalise when monitoring evidence remains in the customer-controlled cloud boundary
Anthropic has announced Enterprise Frontier Safeguards, combining zero data retention with automated detection of serious misuse. The service is designed to keep monitoring data in cloud infrastructure controlled by the customer, with customer-managed encryption keys and automated review available as opt-in controls. Anthropic says rollout will begin in phases later this fall across Claude products and supported cloud platforms.
Enterprise safeguard announcement
Anthropic
Open edition
Agents · September 3, 2026
Agent governance needs a complete inventory before teams can judge ownership, permissions, and blast radius
Okta has added AI-agent discovery to all existing Identity Security Posture Management subscriptions. Its inventory can surface endpoint agents and MCP servers, builder-platform agents, OAuth grants, and shadow AI applications, then place agent-related misconfigurations alongside other identity risks. Some discovery sources remain Early Access or planned, so coverage should be validated against the organisation's actual endpoints and builder platforms.
Identity Security Posture Management product announcement
Okta
Open edition
Security · September 2, 2026
Agentic operations need a governed execution model when analysis can turn directly into enterprise action
Palo Alto Networks has acquired Console, an AI-native platform designed for agentic workflows across enterprise operations. Palo Alto Networks says Console will deepen Cortex capabilities that help teams investigate signals, prioritise work, and act across their environment; the announcement describes natural-language workflows that can automate alerts, issues, and requests.
Agentic operations acquisition announcement
Palo Alto Networks
Open edition
Business · September 2, 2026
AI service governance needs a live risk-and-accountability plan when regulatory oversight reaches the platform layer
The European Commission has designated ChatGPT a Very Large Online Search Engine under the Digital Services Act after the service declared at least 45 million average monthly EU users. The designation gives OpenAI four months, through January 2027, to meet additional obligations covering systemic-risk assessment and mitigation, transparency, and independent audits.
Digital Services Act designation notice
European Commission
Open edition
Security · September 1, 2026
AI gateways become operationally useful when model access, tools, and network permissions can be governed through one identity-aware control point
Tailscale has made Aperture generally available for securing and managing AI agents and LLM sessions. Its documented controls include model access and spend controls, request and response hooks, guardrails, logging, MCP support, API proxying, and approval-based access to tailnet nodes and SSH resources.
AI gateway general-availability announcement
Tailscale
Open edition
Tools · September 1, 2026
Coding-agent rollouts need a migration path when a model provider can change the terms of a familiar developer tool
OpenAI says it plans to wind down its contract providing OpenAI models to Cursor and has proposed a transition period through November 12, 2026; Cursor could choose to end access sooner. OpenAI documents two continuation paths inside Cursor: bring an API key for local Chat and Agent features, or use the Codex IDE extension with a ChatGPT subscription or API key.
Official product-transition guidance
OpenAI
Open edition