AI Insights
Curated AI news, tool launches, and market trends — handpicked weekly.
Live AI Tools
Free tools built and maintained by NaSy Hub
Latest AI News
Top stories from the past week
Anthropic Reduces Claude Fable 5 Access for Paid Subscribers Amid Competitive Pressure
Anthropic announced significant changes to Claude Fable 5 access plans effective July 20. Claude Max and Team Premium subscribers face 50% usage limit reductions, while Pro and Team Standard subscribers lose access under current plans, requiring API pricing with a one-time $100 credit. The changes are widely seen as a response to competitive pressure from OpenAI's GPT-5.6 Sol.
Google Gemini 3.5 Pro Misses Third Release Target as Coding Issues Persist
Google's Gemini 3.5 Pro has missed its third release target, failing to meet expectations in coding and complex reasoning during internal testing. Announced at Google I/O for a June release, the launch has now slipped past July 17. Reports suggest the delay stems from code generation challenges and internal organisational issues, with Google now reportedly considering a stopgap Gemini 3.6 Flash release.
EU Orders Google to Open Android to Rival AI Assistants Under Landmark DMA Decision
The European Commission issued binding decisions under the Digital Markets Act ordering Google to grant rival AI assistants the same Android access as Gemini — including microphone, camera, wake-word triggers and screen automation. Google must also share anonymised Search data with competitors from January 2027. Android 18 compliance is required by August 2027, marking the most significant AI regulation intervention to date.
Anthropic Begins IPO Investor Meetings at $965 Billion Valuation
Anthropic has started meetings with prospective investors ahead of a potential October 2026 IPO. Goldman Sachs, Morgan Stanley and JPMorgan Chase are lead underwriters. The company was last valued at $965 billion after its Series H round in May 2026, positioning it as one of the most valuable AI-native companies to go public — potentially ahead of competitor OpenAI, which has reportedly pushed its IPO timeline to 2027.
OpenAI Details GPT-Red: Automated Red-Teaming Model That Beats Humans at Prompt Injection
OpenAI unveiled GPT-Red, an internal automated red-teaming model that uses self-play reinforcement learning to discover prompt injection vulnerabilities. GPT-Red achieved an 84% success rate finding indirect prompt injections versus 13% for human red-teamers. Insights are integrated into GPT-5.6 Sol, which now shows six times fewer prompt injection failures than GPT-5.5.
200+ Economists and Tech Leaders Warn AI Could Cause 'Large-Scale Job Displacement'
Over 200 economists, researchers and tech leaders including Nobel laureates and executives from Anthropic, OpenAI and Google issued a joint statement calling for guardrails on AI. The 'We Must Act Now' initiative warns AI could drive an unprecedented economic transformation with both major gains in living standards and risks of large-scale job displacement.
MIT Researchers Develop Foundation Model for Tabular Time-Series Data via Ikigai Labs
MIT Professor Devavrat Shah's lab developed a foundation model for tabular time series data, commercialised through spin-off Ikigai Labs. Unlike LLMs trained on text and images, this model takes structured row-and-column data as input to provide real-time forecasting, planning and decision-making for enterprises.
Elon Musk Showcases Prototype AI Hardware Device to SpaceX Investors
Elon Musk demonstrated a prototype of a new consumer AI hardware device to SpaceX investors, signalling his expansion into AI hardware beyond Tesla and xAI. Details of the device remain limited, but the move suggests Musk is entering the consumer AI device market.
Cloudflare Launches Monetization Gateway for AI Agent Payments via x402 Protocol
Cloudflare announced the waitlist for its Monetization Gateway, built on the x402 protocol. The new infrastructure automates online payments for digital services, including those powered by AI agents, enabling seamless microtransactions between AI systems and service providers.
Accenture and Google Cloud Launch Pre-Built Agentic AI Solutions for Mid-Market Companies
Accenture Edge and Google Cloud partnered to launch a suite of pre-built agentic AI solutions targeting mid-market companies. The offerings leverage Google's AI stack for customer intelligence, cybersecurity, and operations — signalling the shift from custom AI builds to packaged enterprise deployments.
Meta Launches Muse Spark 1.1, a Multimodal Reasoning Model for Agentic Tasks
Meta introduced Muse Spark 1.1, a multimodal reasoning model designed for agentic tasks with a 1-million-token context window, supporting text, images, video, audio and PDF inputs. The model can orchestrate multi-agent workflows, shows zero-shot generalisation to new tools and MCP servers, and is competitive with GPT-5.5 on benchmarks. A public preview of the Meta Model API was also released.
Anthropic Launches Claude Science and Expands Project Glasswing Cybersecurity Initiative
Anthropic introduced Claude Science, an AI workbench integrating scientific databases and computational tools for drug research. The company also expanded Project Glasswing — a cybersecurity initiative using the Claude Mythos model — to 150 organisations across 15 countries, and launched Claude Corps, a 12-month paid fellowship for AI professionals in non-profits.
OpenAI Launches GPT-Live: Full-Duplex Voice AI for Real-Time Conversations
OpenAI unveiled GPT-Live, a next-generation voice AI built on full-duplex architecture that listens and speaks simultaneously. The model powers ChatGPT Voice with live translation, web search, and task delegation to frontier models like GPT-5.5. Two versions rolled out globally: GPT-Live-1 for paid users, GPT-Live-1 mini for Free tier.
LLM Model Watch
Track the latest frontier models and their status
Market Pulse
Key signals and trends shaping the AI landscape
AI Market
$1.3T by 2030
Top Skill
Agent Orchestration
Hot Role
AI Solutions Engineer
Get AI Briefing in Your Inbox
Weekly curated AI news, LLM releases, and market trends. Zero spam. Unsubscribe anytime.