AI News

Latest updates from the world of artificial intelligence

MODEL WATCH · OPENAI · LEAKS, UNCONFIRMED JUN 17 GPT-5.6 looks days away — if the leaks hold. Prediction markets put a late-June launch near 83% — but OpenAI has confirmed nothing. 83% LAUNCH ODDS Polymarket · June 22–28 window RUMORED SPECS Context window up to ~1.5M tokens Stronger agentic coding, a leaker claims Pricing rumored near a third of Fable 5 BITSMINDS.COM Source: Polymarket · Cryptopolitan · leaks
Models

GPT-5.6 Rumors Reach a Fever Pitch: Prediction Markets Bet on a Late-June Launch, Leaks Claim a 1.5M-Token Window

No OpenAI announcement exists yet, but developer leaks and prediction markets now point to a GPT-5.6 launch in late June — Polymarket prices the June 22–28 window near 83% — with rumored upgrades including a ~1.5M-token context and stronger agentic coding. Treat it all as unconfirmed.

Cryptopolitan

Read more →
AI SAFETY · OPENAI JUN 16 OpenAI now rehearses a model before it ships. Deployment Simulation replays 1.3M past chats through a new model to forecast misbehavior. STEP 1 Replay recent production chats STEP 2 Regenerate the reply, new model STEP 3 Grade it for misbehavior STEP 4 Estimate the deployment rate Median error 1.5x · caught “calculator hacking” in GPT-5.1 before release Built from ~1.3M de-identified conversations, Aug 2025 to Mar 2026. BITSMINDS.COM Source: OpenAI · MarkTechPost
Research

OpenAI’s “Deployment Simulation” Replays 1.3 Million Real Chats to Catch a Model Misbehaving Before Launch

Detailed June 16, OpenAI’s Deployment Simulation forecasts how an unreleased model will behave by replaying ~1.3M past conversations through it and grading the results — hitting a median error of about 1.5x and catching “calculator hacking” in GPT-5.1 before release.

OpenAI

Read more →
PIXEL DROP · JUNE 2026 JUN 17 Gemini moves into the camera roll. The June Pixel Drop ships AI video and music creation, plus stable Android 17. Gemini Omni Generate video from a text prompt or your own photos. Lyria music Make original music tracks from text or an image. Screen Reactions Selfie green-screen baked into screen recordings. ALSO IN THE DROP · ANDROID 17 BUBBLES · VOICE TRANSLATE ON PIXEL 10a · WEAR OS 7 BITSMINDS.COM Source: Google · The Keyword
Products

Google’s June Pixel Drop Puts Gemini Video and Music Creation on the Phone, Alongside Stable Android 17

Rolling out June 17, the June 2026 Pixel Drop adds Gemini Omni for AI video, Lyria for AI music, and a Pixel-first Screen Reactions tool — and ships stable Android 17 with Bubbles floating-window multitasking and Voice Translate on the Pixel 10a.

Google

Read more →
ZHIPU / Z.AI · OPEN MODEL JUN 16 GLM-5.2 goes open under MIT. A 744B-parameter MoE built for long-horizon agentic coding. 744B TOTAL · 40B ACTIVE (MoE) 1M CONTEXT TOKENS 131K MAX OUTPUT TOKENS MIT OPEN-WEIGHTS LICENSE No benchmarks published at launch — performance is a vendor claim. glm-5.2[1m] · 28.5T training tokens · agentic coding BITSMINDS.COM Source: Z.ai
Models

Zhipu Releases GLM-5.2: a 744B-Parameter, 1M-Token Coding Model Under a Full MIT License

Zhipu (Z.ai) has launched GLM-5.2, a 744B-parameter Mixture-of-Experts model (40B active) with a 1M-token context window, built for agentic coding and released under a permissive MIT license. The catch: no benchmarks were published at launch, so performance claims remain vendor assertions for now.

Z.ai

Read more →
AI ASSISTANTS · MOBILE MARKET SHARE JUN 16 ChatGPT slips below 50% — a first. Sensor Tower: 46.4% ChatGPT, 27.7% Gemini, 10.3% Claude by late May. 46.4% CHATGPT ChatGPT 46.4% Gemini 27.7% Claude 10.3% DOWN FROM 81% TWO YEARS AGO · 1.1B / 662M / 245M MONTHLY USERS BITSMINDS.COM Source: Sensor Tower · TechCrunch
Industry

ChatGPT Slips Below 50% Market Share for the First Time as Gemini and Claude Close In

Sensor Tower data reported June 16 puts ChatGPT at 46.4% of the mobile AI-assistant market by late May — below half for the first time, down from about 81% two years ago — as Gemini climbs to 27.7% and Claude to 10.3%. Claude now also out-earns ChatGPT on revenue per user.

TechCrunch

Read more →
ANTHROPIC · CLAUDE AGENT SDK JUN 16 Nothing changes — for now. Anthropic pauses its June 15 billing overhaul amid the OpenAI price war. INCLUDED API RATES $0 EXTRA — STILL INCLUDED AGENT SDK · claude -p · GITHUB ACTIONS · THIRD-PARTY AGENTS STAY ON YOUR PLAN BITSMINDS.COM Source: The Decoder
Companies

Anthropic Backs Off Its Claude Agent SDK Billing Overhaul on the Day It Was Due — “Nothing Changes for Now”

On the day it was set to take effect, Anthropic paused its plan to move Claude Agent SDK, claude -p, GitHub Actions and third-party agent usage off subscriptions into a metered credit pool. Usage stays on regular Pro and Max plans “for now” — a retreat shaped by developer backlash, a looming OpenAI price war and an approaching IPO.

The Decoder

Read more →
SPACEX · ACQUISITION JUN 16 SpaceX buys Cursor for $60B. An all-stock merger to hand xAI a coding powerhouse. $60B ALL-STOCK Q3 2026 EXPECTED CLOSE </> VIA X67 SUBSIDIARY · NO IPO CASH · $4B ANTITRUST BREAK FEE · CURSOR ARR ~$4B BITSMINDS.COM Source: CNBC · Yahoo Finance
Companies

SpaceX Signs the $60B Deal to Buy Cursor — an All-Stock Merger That Hands xAI a Coding Powerhouse

SpaceX confirmed a definitive all-stock agreement on June 16 to acquire Anysphere, the maker of Cursor, at a $60 billion valuation — routed through a subsidiary so no IPO cash is touched. The deal hands Musk’s xAI a top AI-coding tool, is set to close in Q3, and carries a $4 billion antitrust break fee.

Yahoo Finance

Read more →
OPENAI · AUDITED 2025 FINANCIALS JUN 16 A $34B spend. $13B in revenue. The audited 2025 books, on the eve of a $1 trillion IPO. $13B $34B REVENUE SPEND $21B gap $19B R&D · ~$8B OPERATING LOSS · $30B NON-CASH CHARGE · IPO UP TO $1T BITSMINDS.COM Source: Financial Times
Companies

OpenAI’s 2025 Books: a $34B Spend, $13B in Revenue and a $39B Loss as the $1T IPO Nears

Audited figures reported by the Financial Times show OpenAI spent about $34 billion in 2025 — roughly $19 billion on R&D alone — against $13 billion in revenue, while a one-off accounting charge pushed its net loss near $39 billion, just as the company files to go public at up to $1 trillion.

Reuters

Read more →
OPEN LETTER TO COMMERCE · 76 SIGNATORIES JUN 15 76 defenders sign on to reverse the ban. An open letter to Commerce asks Washington to restore Fable 5 and Mythos 5. “You’ve taken the best models away from the defenders.” — 76 cybersecurity veterans, incl. Stamos · Moussouris · Vixie STILL OFFLINE · NO TIMELINE BITSMINDS.COM Source: TechCrunch · Cybersecurity Dive
Industry

Disarming the Defenders: 76 Security Veterans Demand Washington Reverse the Fable 5 Ban

An open letter signed by 76 cybersecurity veterans — among them Alex Stamos, Katie Moussouris and Paul Vixie — urges Commerce Secretary Howard Lutnick to lift the export-control order that pulled Claude Fable 5 and Mythos 5 offline, arguing it strips defenders of their best tool while doing nothing to slow adversaries. Anthropic warns the precedent could “halt all new model deployments” across the industry.

TechCrunch

Read more →
AI DEBT BOOM · INVESTMENT-GRADE BONDS JUN 15 Even Nvidia is borrowing now. A $20B bond sale across seven tranches — its first since 2021. 2Y 3Y 5Y 7Y 10Y 20Y 30Y $20B TARGET · 7 TRANCHES · 2–30 YEARS · 4× THE 2021 SALE BITSMINDS.COM Source: Bloomberg
Industry

Even Nvidia Is Borrowing Now: Chipmaker Targets $20B in Its First Bond Sale Since 2021

Nvidia, one of the most cash-rich companies on earth, is tapping the bond market for at least $20 billion across seven tranches maturing in two to thirty years — its first investment-grade issuance since 2021. The move puts the AI boom’s biggest beneficiary squarely inside the debt wave now funding the industry’s build-out.

Bloomberg

Read more →
Sarvam Becomes India’s Newest AI Unicorn With a $234M Round Led by HCLTech
Companies

Sarvam Becomes India’s Newest AI Unicorn With a $234M Round Led by HCLTech

Bengaluru’s Sarvam AI has raised $234 million at a $1.5 billion valuation, joining India’s unicorn club on the strength of foundation models built for Indian languages. IT services giant HCLTech led with $150 million, and the cash will fund Sarvam’s next frontier model and a push into agentic, coding and cybersecurity workloads.

TechCrunch

Read more →
G7 · ÉVIAN-LES-BAINS · 15–17 JUNE 2026 Three stacks, no shared rulebook Washington resists binding rules; Europe and Canada build sovereign stacks of their own NO COMMON G7 RULEBOOK UNITED STATES "American AI stack" EUROPEAN UNION Tech Sovereignty package CANADA National AI strategy BITSMINDS.COM Source: Tech Policy Press
Industry

At the G7, Allies Split Over Who Governs AI — and Whether Anyone Should

The summit photo-op hides a deeper fight: Washington resists binding multilateral AI rules to protect its lead, while Europe and Canada race to build "sovereign" stacks of their own. Expect the governance language at Évian to be watered down.

Tech Policy Press

Read more →
G7 SUMMIT · ÉVIAN-LES-BAINS, FRANCE · JUNE 15–17 Three rivals, one table The first time all three frontier-lab chiefs face world leaders together OpenAI Sam Altman Anthropic Dario Amodei Google DeepMind Demis Hassabis G7 · 7 NATIONS + THE EU · HOST: FRANCE BITSMINDS.COM Sources: The Next Web, Bloomberg
Industry

Altman, Amodei and Hassabis Land in France Together as the G7 Puts AI Governance Center Stage

For the first time, the chiefs of OpenAI, Anthropic and Google DeepMind sit before G7 leaders at once — at a three-day summit in Évian-les-Bains where youth safety, frontier risks and a package of voluntary commitments top the agenda.

The Next Web

Read more →
ANATOMY OF A TAKEDOWN How a rival’s late-night call darkened Fable 5 Amazon is Anthropic’s top investor, a board backer, and its cloud host. THURSDAY EVE Amazon’s Jassy alerts Treasury WHITE HOUSE Cyber chief meets; Anthropic refuses 5:20 PM ET Export order + 90-min deadline BY 10 PM ET Both models go dark, worldwide BITSMINDS.COM Source: Axios · The Information
Companies

Amazon Led the Push That Took Anthropic’s Fable 5 Offline, Reports Say — Its Own Investor and Cloud Host

New reporting from Axios and The Information says Amazon CEO Andy Jassy personally alerted the Treasury secretary to a claimed Fable 5 jailbreak, and at least five other companies piled on — setting off the export-control order that darkened Anthropic’s flagship within hours.

Axios

Read more →
Moonshot AI Ships Kimi K2.7-Code, an Open-Weight Coding Model It Says Uses 30% Fewer Reasoning Tokens
Models

Moonshot AI Ships Kimi K2.7-Code, an Open-Weight Coding Model It Says Uses 30% Fewer Reasoning Tokens

The Beijing lab’s fifth K2 release in a year is a coding-first, 1-trillion-parameter open-weight model under a modified MIT license — but independent testers say the headline benchmarks are Moonshot’s own and don’t all hold up.

Crypto Briefing

Read more →
Humanoid Robot Maker EngineAI Files for a Hong Kong IPO as China’s Robot Listing Wave Hits $22.6B
Industry

Humanoid Robot Maker EngineAI Files for a Hong Kong IPO as China’s Robot Listing Wave Hits $22.6B

Shenzhen-based EngineAI, valued at $1.5 billion just three years after its founding, has filed confidentially to list in Hong Kong — the latest entrant in a humanoid-robot IPO rush that has already raised $22.6 billion this year.

The Next Web

Read more →
RAMP AI INDEX · APRIL 2026 Anthropic passes OpenAI in U.S. business adoption Anthropic OpenAI 40% 30% 20% 10% 0% 34.4% 32.3% Apr 2025 Oct 2025 Apr 2026 BITSMINDS.COM Source: Ramp AI Index · share of U.S. businesses
Companies

Anthropic Overtakes OpenAI in U.S. Business Adoption for the First Time, Ramp Data Shows

More American companies are now paying for Claude than for ChatGPT, according to the Ramp AI Index — the first crossover since the generative-AI race began, capping a year in which Anthropic quadrupled its enterprise footprint.

Ramp AI Index

Read more →
CLAUDE CODE · TERMINAL + DESKTOP FIVE WAYS TO RUN Permission Modes Shift+Tab in the terminal · Ctrl+Shift+M in the app 1Ask permissionsprompts before edits and commands 2Accept editsauto-approves file edits in scope 3Plan moderead-only; plans, never edits 4Auto moderuns with a background safety check 5 Bypass permissions skips every prompt — sandbox only NEW BITSMINDS.COM Source: Claude Code docs
Products

Claude Code’s Five Permission Modes, Explained: Ask, Accept Edits, Plan, Auto, and Bypass

Claude Code’s desktop app now packs five permission modes into one Ctrl+Shift+M menu — Ask permissions, Accept edits, Plan, Auto, and the newly added Bypass permissions. Here’s what each one does and when to reach for it.

Claude Code Docs

Read more →
OpenAI Acquires Ona to Run Codex Agents Inside the Customer’s Own Cloud
Companies

OpenAI Acquires Ona to Run Codex Agents Inside the Customer’s Own Cloud

OpenAI is buying Ona — the German cloud startup formerly known as Gitpod — to give its Codex agents secure, persistent environments that run inside a company’s own infrastructure, a direct answer to enterprise fears about turning autonomous coders loose.

The Next Web

Read more →
U.S. EXPORT-CONTROL DIRECTIVE Fable 5 & Mythos 5 Pulled Offline Anthropic complies, disabling both for everyone BITSMINDS.COM
Companies

Anthropic Pulls Fable 5 and Mythos 5 Worldwide to Comply With a U.S. Export-Control Directive

Three days after launch, Anthropic disabled Claude Fable 5 and Mythos 5 for every customer to comply with a Commerce Department order that bars foreign-national access on national-security grounds. The company calls it a misunderstanding over a "minor" jailbreak and is fighting to restore access.

Anthropic

Read more →
BITSMINDS.COM
ModelsLab

Fable 5 vs Opus 4.8, One Prompt Each: Four Builds, Dead Level at 2–2

We pasted identical prompts into Claude Opus 4.8 and Claude Fable 5 — an animated highway interchange, a one-file Space Invaders, a pure-SVG aquarium, and a 3D rocket launch — one shot each, no retries. Everything runs live inside the article. Four builds in, it is dead level at 2–2 — and the launch round flips the very pattern the first three revealed.

BitsMinds Lab

Read more →
SOVEREIGN AI · KOREA NAVER Scales to Gigawatt on NVIDIA DSX Sovereign AI factories at GAK Sejong, ramping 55 MW → gigawatt 55 MW H1 2027 100 MW H2 2027 200 MW 2028 1 GW+ GIGAWATT GOAL Planned operating-capacity ramp at GAK Sejong — source: NVIDIA & NAVER BITSMINDS.COM
Industry

NAVER Goes Gigawatt: A Sovereign-AI Factory Build on NVIDIA DSX, Starting at 55 MW

NVIDIA and NAVER will build sovereign AI factories on the DSX platform at the GAK Sejong data center — 55 MW in 2027, ramping toward a gigawatt — to run HyperCLOVA X, a Seoul world model, and an agentic AI layer for Korea and beyond.

NVIDIA Newsroom

Read more →
CLOUD PARTNERSHIP OpenAI Comes to Oracle Cloud Frontier models and Codex, drawable against your OCI commitment Universal Credits ORACLE OCI Oracle Cloud INFRASTRUCTURE </> Models + Codex OPENAI Eligible Oracle Universal Credits now apply to OpenAI usage on OCI BITSMINDS.COM
Products

OpenAI Comes to Oracle Cloud: Customers Can Spend Universal Credits on GPT Models and Codex

Oracle customers will soon be able to apply their existing Universal Credits toward OpenAI models and the Codex coding agent through OCI — folding frontier AI into the cloud commitments enterprises have already signed.

OpenAI

Read more →
OpenAI Weighs Steep Token Price Cuts to Fend Off Anthropic as the Pre-IPO Token War Escalates
Industry

OpenAI Weighs Steep Token Price Cuts to Fend Off Anthropic as the Pre-IPO Token War Escalates

A Wall Street Journal report says OpenAI is weighing significant cuts to what it charges per token, bracing for the same move from Anthropic — a volume play between two rivals both racing toward the public markets.

The Wall Street Journal

Read more →