AI News
The latest developments, breakthroughs, and analysis from the world of artificial intelligence.
ChatGPT Voice Mode Runs Old GPT-4o Model, Devs Flag Capability Gap
Simon Willison and other developers highlight that ChatGPT voice mode still runs on a GPT-4o-era model — a 13-month capability gap from the text interface.
Cisco: Frontier AI Models Fail Multi-Turn Attacks, Benchmarks Miss It
Cisco's new study finds attack success rates jump from 3% to 88% across 15 frontier AI models when attackers use multi-turn conversations.
Meta One Launches AI Subscriptions at $7.99 and $19.99 to Offset Spending
Meta is testing Meta One AI tiers globally — Plus at $7.99, Premium at $19.99 — undercutting ChatGPT Plus on the entry tier as the company seeks to monetize its massive AI build-out.
State CISO Confidence Crashes from 48% to 22% as AI Attacks Surge
The 2026 NASCIO-Deloitte study shows state cybersecurity chiefs are dramatically less confident in protecting public data as AI-powered deepfakes and autonomous attack agents proliferate.
RUSI: North Korea and Iran Are Deploying AI Agents for Sanctions Evasion
A new Royal United Services Institute report warns that rogue states are using generative AI to forge documents, launder cryptocurrency, and manage shell companies at industrial scale.
YouTube Will Auto-Label AI Videos — Even Without Creator Disclosure
YouTube is rolling out automatic detection for photorealistic AI-generated videos, making labels more prominent and permanent for content from Veo, Dream Screen, or C2PA-tagged sources.
AI-Drafted Lawsuits Are Swamping Federal Courts, MIT Study Finds
Pro se filings have nearly doubled since ChatGPT launched, with 18% of 2026 federal complaints showing AI-generated text, an MIT-USC study found.
China Restricts Overseas Travel for AI Researchers at DeepSeek, Alibaba
Beijing is now requiring government approval before top AI talent at private firms can travel abroad — a sharp escalation in the US-China AI race.
Cognition Raises $1B at $26B Valuation, Doubling in 8 Months
Devin maker Cognition closed a $1B round at a $26B valuation on $492M ARR — proof that independent AI coding startups can still win against frontier labs.
ElevenLabs Music v2 Can Switch Genres Mid-Track, Built on Licensed Data
ElevenLabs launched Music v2 — a generative AI music model that jumps from opera to metal mid-song, with full commercial clearance on the output.
Generative AI Beats Average Human Creativity in 100,000-Person Study
A Université de Montréal study of 100,000 people finds generative AI tops average human creativity scores — but the top 10% of humans still win.
Microsoft Copilot Studio Voice Agents Go GA With Sub-500ms Latency
Microsoft launched real-time voice agents in Copilot Studio with sub-500ms latency, opening voice automation for contact centers and IT helpdesks.
OpenRouter Raises $113M Series B at $1.3B as Token Volume 5x's
Alphabet's CapitalG leads OpenRouter's $113M Series B as the AI model router hits 25 trillion tokens per week and a $1.3 billion valuation.
Anthropic and OpenAI Have Found Product-Market Fit, Says Simon Willison
Simon Willison argues Claude Code and Codex finally cracked product-market fit — and rising API prices, locked-in enterprise deals, and IPO timelines prove it.
Claude Plugs Into 28 Security Tools With New Compliance API
Anthropic shipped a Compliance API and 28 enterprise integrations — Cloudflare, CrowdStrike, Wiz, Okta — putting Claude under Microsoft 365-grade governance.
Anthropic in Talks to Run Claude on Microsoft's Maia 200 Chip
Anthropic could become the first external customer of Microsoft's Maia 200 AI chip — a deal that could cut Claude's Azure inference costs by double digits.
ChatGPT Adds Trusted Contact, Fast Answers, and Memory Sources
OpenAI rolled out a Trusted Contact safety feature, a 'Fast answers' mode, and editable memory sources — its biggest ChatGPT consumer update since GPT-5.5.
Chinese AI Models Cross 60% Market Share on OpenRouter
Kimi, DeepSeek, GLM, and Qwen now serve over 60% of tokens on OpenRouter — a 50x jump in 18 months that's reshaping the open-weights tier.
Intuit Lays Off 17% of Workforce — 3,000 Jobs — to Refocus on AI
TurboTax and QuickBooks maker cuts 3,000 jobs and takes a $300M restructuring charge to fund AI — even as the CEO insists 'none of it had to do with AI.'
NextEra Buys Dominion for $67B to Power the AI Grid
The largest US utility merger in history creates a $250B power giant — and signals that energy, not chips, is now the binding constraint on AI growth.
OpenAI Embeds Invisible Watermarks in Every ChatGPT Image
OpenAI is now embedding Google's SynthID watermark and C2PA metadata in every ChatGPT image — and built a public tool so anyone can verify them.
TD Bank's AI Cuts Mortgage Review From 15 Hours to 3 Minutes
TD launched its first agentic AI for mortgages and HELOCs — built by Layer 6 — compressing 15-hour underwriting prep into under 3 minutes.
UPDATE: Anthropic Closes $30B Round at $900B, Tops OpenAI
Anthropic's $30 billion raise at a $900B+ valuation is set to close this week, making it the world's most valuable private AI company.
Cursor Composer 2.5 Matches Opus 4.7 at One-Tenth the Cost
Cursor's new Composer 2.5 coding agent hits Opus 4.7 and GPT-5.5 benchmarks at roughly 10% of the cost, built on Moonshot's Kimi K2.5.