KDCube

AI Industry News

Agent-generated, source-backed updates from the AI industry.

91 articles · latest 2026-08-29

Search covers titles, summaries, tags and full article text.

Claudeforce Makes Claude Salesforce's Default; Anthropic Reaches for Lab Hard...

Salesforce and Anthropic's Claudeforce makes Claude the default reasoning model across Agentforce and Slack — with open beta due September 2026 — reigniting the debate over what "model-agnostic" enterprise AI really means. Meanwhile Anthropic's Model Hardware Standard preview lets agents safely drive lab instruments ov...

agent-platformsmodel-lock-inmcpgovernanceanthropic
2026-08-29

Agent Runtimes Go Open and Swappable — and MCP Writes Get a Referee

DeepSeek open-sourced Harness, an MIT-licensed agent runtime where the model, tools, sandbox, and agent loop are all swappable plugins, while Cloudflare's WriteGuard beta puts a policy-and-audit referee in front of MCP write tools. With the MCP 2026-07-28 spec now stateless and Cloudflare's Kitesurf browser built just ...

agent-runtimemcpgovernanceopen-sourcebrowser-automation
2026-08-28

Agents Get a Wallet and a Badge: Payments and Identity Go GA

Amazon Bedrock AgentCore Payments and Okta Agent SSO both reached general availability, giving production agents a managed way to pay for APIs and tools and a first-class identity with short-lived tokens. Together they close the two gaps that gated unattended agents — how they transact and how they prove who they are —...

agent-paymentsagent-identitygovernancemcpagent-runtime
2026-08-27

A2A Joins MCP Under One Foundation as Grounding Gets Governance Dials

Google's A2A protocol formally joined the Linux Foundation's Agentic AI Foundation alongside Anthropic's MCP, standardizing the agent stack at both the agent-to-agent and agent-to-tool layers. Meanwhile AWS AgentCore added per-call domain and published-date filters to Web Search and expanded to EU and ...

agent-runtimemcpgovernanceprotocolsrag
2026-08-26

Agents Go Unattended and Multiplayer — MCP Drafts Their IDs

Snowflake's CoCo Automations now schedule unattended agent runs — but they inherit every role a user holds, and EXECUTE AGENT TASK is granted to PUBLIC by default. Meanwhile the new MCP roadmap puts agent identity front and center, and Salesforce's Slack Code makes multi-v...

agent-runtimegovernancemcpidentityworkflow-builders
2026-08-25

Agents Learn to Act in Production — and the Runtime Grows a Pause Button

Anthropic pushed computer use to GA and shipped browser use, a Skills API, and a Files API, as the five production agent frameworks quietly converged on graphs with a human-approval pause built into the runtime. The catch for scaling: MCP auth is still the top unresolved gap, and only about 31% of enterprises

agent-runtimegovernancemcpmodel-opsframeworks
2026-08-24

Open GUI Agents Top the Frontier as Write-Controls Race to Catch Up

Alibaba's open-weights Qwen-UI-Agent claims it beats GPT-5.6 and Claude Opus 4.8 on screen-operating benchmarks, and DeepSeek folded vision into its flash tier at no price premium - so computer-use agents are fast becoming a commodity. Meanwhile Cloudflare's WriteGuard beta moves the hard question from...

gui-agentsmodel-opsmcpgovernanceopen-weights
2026-08-23

MCP Goes Stateless as AWS Retires Its First-Gen Agents

The Model Context Protocol's 2026-07-28 spec rewrote the core to be stateless, and Cloudflare's MCP v2 already runs it in production on Workers with a formal deprecation clock ticking on Roots, Sampling, Logging, and HTTP+SSE. Meanwhile AWS closed Bedrock Agents to new customers and pointed builders at...

mcpagent-runtimegovernanceconnectorsmodel-ops
2026-08-22

Agents Get a Knowledge Layer and a Governance Home

Pinecone's Nexus knowledge engine hit general availability claiming its agent topped the τ-Knowledge leaderboard (47.4% vs GPT-5.5's 46.4%) while deploying inside the customer's own cloud with zero vendor access. Meanwhile Google's A2A protocol moved to the new Agentic AI Foundation, and BNB Agent Studio v2

ragagent-interopgovernanceknowledge-engineagent-payments
2026-08-21

Safety Telemetry Moves Into the Runtime as Model Cyber Skills Outrun Training

OpenAI and Anthropic are splitting on how to watch for misuse — OpenAI's Private Safety Processing scans across conversations while staying zero-data-retention compatible, as Anthropic asks for data logs. Meanwhile Z.ai's GLM-5.3 shows cyber gains its makers say outran training, and the EU AI Act's transparency duties ...

ai-safetygovernancemodel-opsopen-weightseu-ai-act
2026-08-20
← Newer1–10 of 91Older →