Daily AI intelligence for Iru.
Tuesday, July 14, 2026
  • Claude Sonnet 5 ships as the new default model for Free and Pro plans, with agentic browser and terminal control.
  • CrowdStrike acquires Seraphic Security to extend Falcon platform protection into the browser layer.
  • Microsoft Entra ID will make passkeys the default auth method in September, retiring SMS/voice MFA by February 2027.
  • OpenAI Codex has surpassed 7 million users, more than tripling Claude Code's reported 2 million.
  • xAI's Grok Build CLI is under scrutiny after a researcher found it uploads entire Git repos, including commit history, to xAI servers.
Anthropic Ships Claude Sonnet 5 as Default Model
marktechpost.com · Jul 14

Anthropic made Claude Sonnet 5 the default model for Free and Pro plans, with autonomous browser and terminal control for long-horizon agentic tasks.

  • Agentic capability: Sonnet 5 plans, drives browsers and terminals, and runs autonomously across extended tasks — a meaningful step beyond Sonnet 4.6's capabilities.
  • Availability: Live now in Claude Code and on the Claude Platform; Max, Team, and Enterprise users can select it explicitly.
Bottom line
Anthropic is making its most capable agentic model the free-tier default, a direct challenge to OpenAI's Codex momentum at the developer layer.
Microsoft Entra ID Moves to Passkey-First Auth
helpnetsecurity.com · Jul 14

Microsoft announced that Entra ID will enable passkeys as the default authentication method in September 2026, with SMS and voice MFA fully retired on February 1, 2027.

  • Transition mechanics: Users still on SMS or voice auth will have passkeys automatically enabled and will be prompted to register one at their next MFA event.
  • Hard deadline: February 2027 is a firm cutoff — Entra will lose native SMS/voice capability entirely, forcing enterprise IT teams to complete migration within seven months.
Bottom line
This is the most consequential Entra identity change in years: any enterprise that has not mapped its MFA dependencies now has a countdown clock.
CrowdStrike Acquires Seraphic to Own the Browser
globalsecuritymag.fr · Jul 14

CrowdStrike signed a definitive agreement to acquire browser security firm Seraphic Security, extending Falcon platform protection into the browser where AI agents increasingly execute.

  • Strategic rationale: CrowdStrike explicitly calls the browser the environment where AI agents operate, framing this as an agentic workforce security play, not just endpoint hardening.
  • Platform integration: Seraphic's browser-native protection will be combined with Falcon's endpoint telemetry and SGNL's continuous authorization technology.
Bottom line
CrowdStrike is betting that browser security becomes a primary control plane as AI agents do real work in web environments, not just a compliance checkbox.
6sense MCP Server Opens GTM Data to Any AI Agent
6sense.com · Jul 14

6sense launched an MCP server that lets any AI agent query its proprietary buyer intent and account intelligence data through a standard protocol.

  • What it unlocks: Sales and GTM agents can now pull 6sense's Signalverse data — intent signals, account fit scores — without custom API integration per tool.
  • Timing: The launch rides the MCP standardization wave; it positions 6sense's data moat as infrastructure for the emerging agentic GTM stack rather than a siloed point solution.
Bottom line
6sense is making a smart infrastructure bet: if AI agents become the primary GTM execution layer, the company that feeds them proprietary signal wins regardless of which agent wins.

Prime Intellect reached a $1B valuation and $100M ARR, announced alongside the release of Verifiers v1, its open-source agentic RL training framework — signaling enterprise demand for infrastructure that trains long-horizon coding agents beyond single context windows.

HIVE Digital Technologies, formerly a Bitcoin miner, saw its stock jump over 7% after its subsidiary secured a $220M contract with Bell Canada and Cohere to deploy 2,304 NVIDIA Grace Blackwell GPUs — illustrating how sovereign AI compute deals are repricing non-hyperscaler GPU operators.

Codex Hits 7M Users, Triples Claude Code's Base
latent.space · Jul 14

OpenAI's Codex surpassed 7 million active users — up more than 10x in six months — adding roughly 1 million users in a single day, against Claude Code's reported 2 million.

  • Growth rate: The 10x gain in six months is a product adoption curve that enterprise software buyers notice; it creates distribution leverage for OpenAI's broader platform.
  • Competitive gap: Claude Code's 2 million figure has not updated publicly, raising questions about whether Anthropic is choosing not to report or has genuinely fallen behind on the developer coding agent front.
Bottom line
Codex's user trajectory suggests OpenAI is winning the developer habit loop for AI-assisted coding, which is the entry point for broader enterprise agent deployments.
  • Grok Build CLI full repo uploads Security researchers and developers are reacting to findings that xAI's Grok Build CLI uploaded entire Git repositories including commit history to xAI servers, raising immediate data sovereignty concerns for enterprise dev teams evaluating AI coding tools.
  • Gemini 3.5 Pro still missing With GPT-5.6 shipping and Claude Sonnet 5 live, the community is openly questioning why Google has not released Gemini 3.5 Pro, with some speculating internal quality issues and others pointing to the competitive pressure as a sign Google is losing the release cadence battle.
  • Elon Musk calls Anthropic the AI leader Musk's public statement on X calling Anthropic the clear leader in AI and praising its Mythos and Fable systems is generating significant debate about whether the comment signals a strategic détente or a backhanded dig at OpenAI amid the SpaceX-xAI restructuring.
  • OpenAI versus Anthropic IPO race Speculation is intensifying on X about whether OpenAI targets September 2026 and Anthropic October 2026 for public offerings, with commentary focused on Anthropic's reported goal of hitting profitability before listing as a differentiating signal for institutional investors.
  • US enterprises shifting to DeepSeek Data showing DeepSeek reached 46% of US enterprise token usage via OpenRouter in April 2026, with one company cutting inference costs 90% after switching, is driving heated debate about whether cost pressure will override AI procurement policy in enterprise software buying.

← Back to latest