AI Daily Brief

Thursday, May 14, 2026

8 stories · Models, enterprise AI, and the agentic wave
Top Story
Anthropic Security · New Model
Breaking

Claude Mythos Preview: Anthropic's Cybersecurity-Focused Model

Anthropic unveiled Claude Mythos Preview, a new model purpose-built to identify software vulnerabilities and security flaws with exceptional accuracy. Due to its power, Anthropic is limiting rollout to a vetted group of companies under a new cybersecurity initiative called Project Glasswing. CEO Dario Amodei personally presented Mythos at the White House to senior members of the Trump administration, signaling both its sensitivity and its strategic importance. The model operates under red.anthropic.com, Anthropic's security research platform.

red.anthropic.com
Models & Releases
OpenAI Model Release

GPT-5.5 Instant Is Now the Default ChatGPT Model

OpenAI has begun rolling out GPT-5.5 Instant as the new default for all ChatGPT users, replacing GPT-5.3 Instant. In internal evaluations, the model produces 52.5% fewer hallucinated claims on high-stakes prompts in medicine, law, and finance. Alongside the rollout, OpenAI introduced three new realtime voice models in the API: GPT-Realtime-2 for live voice reasoning, GPT-Realtime-Translate for multilingual speech, and GPT-Realtime-Whisper for streaming transcription. Enhanced personalization from past chats, files, and connected Gmail is rolling out to Plus and Pro users on web.

openai.com
Google DeepMind Feature · Demo

DeepMind Demos "Magic Pointer" — Gemini Inside Chrome's Cursor

Google DeepMind detailed a new UI feature called Magic Pointer on Googlebook, designed to give the system pointer semantic awareness: the AI understands not just what you are pointing at but why it matters to you. Users will be able to use their pointer to ask Gemini in Chrome about any element on a webpage — comparing products, visualizing objects in a room, or surfacing contextual answers. The feature will ship to Chrome later this year and expand to Gemini Intelligence on Android, starting with Samsung Galaxy and Pixel devices this summer.

9to5google.com
Enterprise & Products
Anthropic Product Launch

Claude for Small Business Brings Agentic Workflows to SMBs

Anthropic launched Claude for Small Business — a bundle of connectors and 15 ready-made agentic workflows targeting finance, operations, sales, marketing, HR, and customer service. It integrates with QuickBooks, PayPal, HubSpot, Canva, DocuSign, Google Workspace, and Microsoft 365. Use cases include invoice chasing, margin analysis, month-end book prep, tax-season organization, contract review, and lead triaging — all runnable without coding.

digit.in
Microsoft Enterprise AI

Microsoft Copilot Cowork: AI Moves From Chat to Execution

Microsoft introduced Copilot Cowork through its Frontier for Business program, extending Copilot beyond conversation into task delegation — accepting meetings, drafting follow-ups, reorganizing calendars, and managing inbox priorities autonomously via natural language commands in Outlook. In an internal analysis of 100,000+ M365 Copilot chats, Microsoft found that 49% of conversations support cognitive work that once required deep expertise, with 58% of users saying they are producing work that was impossible a year ago.

microsoft.com
Policy & Industry
Industry Policy · Security

US Government to Pre-Test AI Models from Google, Microsoft, and xAI

The Center for AI Standards and Innovation (CAISI) under the Commerce Department announced agreements with Google DeepMind, Microsoft, and Elon Musk's xAI to conduct pre-deployment evaluations of frontier AI models. The initiative allows the US government to assess capabilities and run targeted security research before models reach the public. OpenAI separately said it is making its most advanced models available to all vetted government levels to help get ahead of AI-enabled threats.

cnbc.com
Meta Investment · Strategy

Meta Commits $115–135B in AI Capex for 2026, Nearly Doubling Last Year

Meta announced AI capital expenditures of $115–135 billion for 2026 — nearly double its 2025 spending — as the company accelerates infrastructure investment to close the gap with OpenAI and Google. The move comes as OpenAI crossed $25 billion in annualized revenue (with Anthropic approaching $19 billion) and reportedly takes early steps toward a public listing as soon as late 2026. Meta's bet is large-scale compute and open-weight models as a long-term differentiator.

techcrunch.com
Dev Tools
GitHub Pricing · Developer Tools

GitHub Copilot Switches to Usage-Based AI Credits on June 1

GitHub is migrating Copilot to usage-based AI Credits effective June 1, 2026. The premium-request multiplier for Opus has already been raised to 15× this month, with a further jump to 27× already scheduled. Developers running Claude Code, Codex CLI, and DeepSeek TUI in parallel terminals — switching by task class — are reportedly shipping fastest this cycle according to developer surveys, as no single tool dominates every workload category.

thenewstack.io