Today's Headlines
- Google makes Gemini 4 Argon generally available, with API pricing from $2 per million input tokens
- Character.AI files for Chapter 11, with Disney bidding $20 million to license some of its IP
- American Express publishes a framework for proving purchase intent in agent payments
- Anthropic tests whether small models could self-replicate, running a 26B model on 15 cheap TV boxes
Google leads the day. Gemini 4 Argon, the frontier model it announced on September 30, is now rolling out worldwide to paying subscribers and to the paid tier of the Gemini API.
The cost of AI, and who pays it, ran through the rest of the news. Character.AI, the AI character chat service, filed for bankruptcy protection, citing development costs and litigation exposure. American Express published a framework for checking what a customer actually approved when an AI agent pays on their behalf.
Today's Top Three
Google opens Gemini 4 Argon to paying users, with API pricing from $2 per million input tokens
Google made Gemini 4 Argon, the frontier model it announced on September 30, generally available worldwide on October 13 (US time).
Developers and enterprises get it through the paid tier of the Gemini API. Consumers get it in the Gemini app and AI Mode in Search on the paid Google AI Pro and Ultra plans, while the agentic coding environment Google Antigravity is limited to Ultra subscribers.
The introductory API price is $2 per million input tokens and $10 per million output tokens, with cached input at 95% off. Ars Technica reports that the rate holds through November 30 and then rises to $4 and $20.
Google says Argon ranks first on the Vals Index, which measures work in finance, law, tax and coding. Inside Google, teams using it in Antigravity raised agentic code submissions by more than 35% and cut code rollbacks by nearly a third.
Google also published agent-safety results. On the Gray Swan indirect prompt injection benchmark, Argon held attack success to 0.7% across 15 adaptive attempts.
The Gemini app now connects to more than 40 apps, including Instacart and Shopify, so users can hand off tasks such as grocery orders and table bookings.
- Gemini 4 Argon: Built to solve your hardest problems (Google DeepMind)
- Gemini 4 Argon comes to the Gemini app. (Google)
- Gemini 4 Argon is now available for developers. (Google)
- Gemini 4 Argon begins rolling out in Gemini app and Antigravity, but only if you pay (Ars Technica)
Character.AI files for Chapter 11, with Disney bidding $20 million to license some of its IP
Character.AI, the service that lets users chat with AI characters, filed for Chapter 11 bankruptcy protection in the US Bankruptcy Court for the District of Delaware.
According to TechCrunch, the company's court filing lists assets and liabilities each in the range of $50 million to $100 million. The filing says the high cost of building and growing AI entertainment, combined with potential litigation liabilities and regulatory engagement, left the company unable to generate enough cash to support the business in its present form.
Disney is the stalking horse bidder in the court-supervised sale. CNBC and Ars Technica report that the $20 million bid covers a non-exclusive perpetual license to certain Character.AI intellectual property, with transition services provided by former Character.AI employees who joined Disney along with former CEO Karandeep Anand.
CEO Colin Adams said the platform will keep running during the sale, which is expected to close in late November or early December. If no bid allows continued operations, the platform will be wound down.
Ars Technica notes that the company's privacy policy allows user information, including chat logs, to be transferred in a sale of company assets.
- Just 5 years in, interactive chatbot platform Character.AI files for bankruptcy (TechCrunch)
- AI companion startup Character.ai files for Chapter 11 bankruptcy, citing costs and litigation (CNBC)
- Disney offers $20M for Character.AI's scandal-plagued chatbot tech (Ars Technica)
- Character Technologies, Inc. (CourtListener)
American Express publishes a framework that makes AI agents carry proof of purchase intent
American Express published Intent Intelligence, a framework for capturing and authenticating a customer's purchase intent before an AI agent completes a transaction, on October 13.
The customer's approved intent reaches the card issuer before the agent buys, and then serves as a control for the rest of the transaction. Amex released the framework publicly to push for a common industry approach, and offers it through its ACE Developer Kit.
It works in four steps. A Create Intent API records the customer's instructions and constraints, such as the merchant and a maximum budget, and the customer authenticates and signs off.
An Intent ID and a Proof of Intent Token are then issued. The agent uses the token to obtain a scoped digital payment credential that it presents at checkout in place of the actual card number.
The ID and token stay attached to the purchase for fraud checks, authorization decisions and dispute resolution. Amex's Hannah Bozian told TechCrunch that if a customer asks for white Nike sneakers and the agent tries to buy Adidas, Amex could stop the transaction.
- American Express Publishes Intent Intelligence Framework for Trusted Agentic Commerce (American Express)
- Amex introduces a framework to prevent AI agents from making unauthorized purchases (TechCrunch)
Other Developments
Models & APIs
- Google DeepMind also showed Gemini 4 Argon at work in robotics. It scored 87% on the ERQA embodied-reasoning benchmark, and after watching a video of a person playing cornhole it built the simulation code that let Apptronik's Apollo humanoid sink seven consecutive shots. Gemini 4 Argon: Built to solve your hardest problems (Google DeepMind) / Gemini 4 Argon for Robotics (Google for Developers)
- SB Intuitions, a SoftBank subsidiary, opened a free beta of its Japanese-built generative AI, Sarashina, with both chat and API access. The API offers six models, including Sarashina3 mini for text generation and Sarashina3 translate for Japanese-English translation. Sarashina free beta (ITmedia AI+, Japanese)
Products
- Shopify will let shoppers set up their own AI agents to browse and buy from its catalog of more than a billion listings. The access, previously tied to partner integrations such as Meta's Muse, now extends to agents without a partnership. Shopify opens its catalog up to more AI agents (The Verge)
- Google now lets users install personal apps built in AI Studio on their own Android devices through Google Play, starting with users in the US aged 18 and over. The apps stay private and separate from the public store. Google Play now lets anyone build private, vibe-coded apps that run on Android (TechCrunch)
- Meta launched Your algorithm on Facebook, which lets people describe in their own words what they want to see in Feed and Reels. Meta says recommendations respond right away. Your Algorithm, Your Words: New Ways to Shape What You See on Facebook (Meta)
- SpaceXSI launched a single subscription that bundles the Grok assistant, the Cursor coding tool and X Premium. It comes in five tiers, from free up to Ultra at $200 a month, with AI usage caps rising at each level. SpaceXSI bundles X Premium, Grok, and Cursor into one subscription (TechCrunch)
Research
- Anthropic's Frontier Red Team published research suggesting that models of 30 billion parameters and under may be months away from being able to self-replicate on consumer IoT devices such as smart ovens and off-brand Android TV boxes. The team ran Google's Gemma4-26B-A4B across 15 TV boxes costing about $55 each, at 0.39 tokens per second. AI models may be capable of self-replicating without access to a datacenter (Anthropic)
Policy
- A lawsuit filed in US court accuses OpenAI of business practices that prey on children. The complaint alleges that OpenAI used ChatGPT to extract vast amounts of highly sensitive and valuable data from millions of children. Doe v. OpenAi Foundation f/k/a OpenAi Inc. (CourtListener)
Business
- NVIDIA announced new investments in four French startups: Doctolib in healthcare, Entalpic in materials discovery, Genesis AI in robotics and Pigment in enterprise planning. It says its commitments across Europe's technology ecosystem now total nearly $10 billion. NVIDIA Announces New Investments in France's Startups (NVIDIA)
- Pluto Security, which monitors how employees use AI tools and agents, has raised $42 million, including a $34 million Series A. Its CEO says about 60% of the risks it detects come from people who are not developers. Pluto Security raises $42M as AI gives enterprise security teams a new headache (TechCrunch)
- Epic Games has given developers on an unannounced Disney shooter an AI-first mandate to generate code and art with tools such as Anthropic's Claude, The Verge reports, citing Bloomberg. The game has slipped from its original March release window. Epic shifts to 'AI-accelerated development' for upcoming Disney game (The Verge)
Source: Selected by the editors from the AI news inbox (collected October 14, 2026: 37 items, 12 primary and 25 secondary).