Latest Artificial Intelligence News & Dispatches

Microsoft Defender's real-time agent protection inspects tool invocations before they run - session context, tool-call allowlists, credential scoping, anomaly detection, egress control - under a read-write agent security model.

WhatsApp's Scam Alert runs a scam-classification model on the device, never uploads message content, and learns only from differentially private aggregates - the reference implementation of private AI safety at planetary scale.

The European Commission's July 16, 2026 DMA orders require Google to open 11 Android features to rival AI assistants by Android 18 (Aug 1, 2027) and share anonymized search data from January 2027 - or face fines up to 10% of global turnover.

Pathway's BDH-CQ is a 150M-parameter non-transformer reasoning model scoring 29.5% on ARC-AGI-1 at $0.0007 per task - 11x cheaper than GPT-5.6 Luna even after OpenAI's 80% price cut.

Ramp's 2026 data splits adoption from spend: Anthropic wins 43.5% of US business clients yet only 6% of their token dollars touch frontier Claude Fable 5, while GPT-5.6 Sol carries 25% of OpenAI's token usage.

BNB Agent Studio v2 turns AI agents into paid workers: ERC-8183 carries the flow from work to wallet settlement, Altana enforces onchain spending limits and allowlists, and a single overnight agent earned $0.31 across nine tasks.

Build a TypeScript FastMCP server that lets AI agents earn via ERC-8183, pay via x402, and spend only inside Altana on-chain limits with Paymaster gas on BSC Testnet.

Build a Python FastMCP server that turns KTX market data, AI signals, portfolio tools, order execution, and prediction markets into MCP tools for live agents.

Microsoft's real-time agent protection evaluates the agentic loop before actions run; wrap it in a LangGraph pre-tool gate that enforces allowlists, detects anomalies, blocks, and auto-remediates.

QF-Test 11.0.1 exposes its automation engine as an MCP server, so Claude Code can discover tests, run suites, analyze failures, and gate releases through one LangGraph loop.

BNB Agent Studio v2 lets agents get hired and paid onchain; wrap the ERC-8183 hire-to-settlement flow in a LangGraph state machine with a human approval gate and Altana wallet limits.

OpenClaw gives agents live browser, terminal, and desktop control; LangGraph gives them state. Here is a supervisor-plus-workers reference implementation with typed retries.

The 2026-07-28 MCP spec removed the session; Enterprise-Managed Authorization lets your IdP authorize agents with no consent popups. This is the LangGraph reference implementation.

Veo 3.1 and Lyria 3.5 are callable APIs, which makes a script-to-screen media pipeline an agent problem. Here is the LangGraph implementation with parallel generation and a review gate.

The MCP 2026-07-28 spec is stateless and the Enterprise-Managed Authorization extension is stable, so your IdP can authorize agents with zero per-user browser consent. This dispatch builds a TypeScript FastMCP server wired to Okta, Entra ID, or Keycloak.