Google ships Gemini 4 Argon to paying users, Character.AI heads toward bankruptcy, and Amazon's million-robot fleet and Amex's agent purchase controls show AI moving deeper into operations and payments.
-
Google releases Gemini 4 Argon for paying users
- Google launched Gemini 4 Argon today for developers and enterprises through the API, and for paying subscribers in the Gemini app and AI Mode in Search.
- The introductory price is $2 per million input tokens and $10 per million output tokens, and it doubles at the end of November.
- Google says Argon ranks first on the Vals Index and scores 87% on the ERQA embodied reasoning benchmark.
- According to The Verge, Argon is Google's first major frontier release since Gemini 3 in late 2025.
Why it matters. The launch puts Google back in the frontier race against GPT-6 Astra and Claude models, which Google says Argon outperforms on several benchmarks. For developers, access is gated by subscription tier, and Antigravity is limited to Ultra subscribers, so plans should account for that before building on the model.
-
Character.AI files for bankruptcy five years after founding
- Character.AI, which lets users create AI characters and chat with them, has filed for Chapter 11 bankruptcy.
- The filing follows a slew of lawsuits and struggles to scale its popular platform.
- The company said it had assets and liabilities between $50 million and $100 million.
- It agreed to sell some assets to Disney for $20 million, along with a non-exclusive perpetual license for its technology.
Why it matters. The filing shows how hard consumer-facing AI products are to scale and monetize. Higher usage also raises AI costs, which adds pressure to charge users more.
-
Epic Games shifts its Disney shooter to AI-first development
- Bloomberg reports that Epic Games is moving to an AI-first development model for its Marvel and Star Wars shooter with Disney.
- Developers were told there would be no more hand-written code and that all art would be generated with full AI workflows.
- The game was slated for March 2027 and is now delayed indefinitely amid the AI shakeup and slower-than-expected development.
- Epic CEO Tim Sweeney said the aim is to build more and better stuff, not to do the same stuff with fewer people.
Why it matters. The project shows a major studio betting that AI models can write most of the code and art for a large commercial game. For developers, it signals AI workflows moving into full production, where people manage model output instead of writing code. A survey cited by Ars Technica found that most AI use in game studios is still for research and brainstorming, not code.
-
Open AI models: benefits weighed against risks
- Anthropic released a report arguing that GLM-5.3 poses a major cyberattack risk.
- The model was released in August by the Chinese company Z.ai and is freely available to download and modify.
- As of publication, the top eight models on the Artificial Analysis LLM leaderboard are all closed, though open models are only a handful of months behind.
- Open models let researchers study their inner workings, and businesses can run them on their own hardware to keep sensitive data secure.
- Once someone downloads an open model, there is no filter checking their prompts and no way to revoke their access.
Why it matters. Open models sit at the center of the debate over who controls frontier AI. Supporters argue they spread power beyond a few companies, while critics warn they are easier to misuse. Any policy involves trade-offs between safety and democratization.
-
Google Play lets users launch private vibe-coded apps on Android
- Google Play now lets users run vibe-coded apps on Android without publishing them to the Play Store.
- Users can build an app with Google AI Studio and launch it through Google Play with a standard Google account.
- These personal apps do not appear in Play search or listing pages and are shared only with their creator.
- Google plans to let users share apps with a small group of friends and family, and to add a path to publish an app later.
Why it matters. Developers who build apps with AI now have a direct way to test them on Android devices. Google is positioning the feature as a route to a professional developer account, with the same Trust & Safety reviews that apply to published apps.
-
Amazon runs over 1 million robots alongside its human workers
- Amazon has more than 1 million robots working in its warehouses.
- Amazon is also the second-largest corporate employer, with more than 1.5 million full-time and part-time human workers, according to its 2025 annual report.
- Amazon says the robots take on the toughest fulfillment center tasks, but labor groups and independent researchers point to evidence that warehouse automation driven by AI is not all upside for workers.
- Amazon CFO Brian Olsavsky said in July that the company expects to more than double its existing fleet of robotic arms this year.
Why it matters. The story shows how AI models now coordinate large robot fleets. Amazon uses a suite called DeepFleet to predict robot congestion, and machine learning raised the success rate of robotic arm picks in one trial. For developers, it is a large-scale case of AI systems working in physical environments next to people.
-
Amex unveils a framework to stop AI agents from making unauthorized purchases
- American Express is introducing the Intent Intelligence Framework, designed to capture and authenticate a customer's purchase intent before an AI agent completes a transaction.
- The framework begins with a Create Intent API that records the customer's instructions and constraints, such as merchant name, maximum budget, and product preferences.
- The customer authenticates their identity and signs off on the captured intent, and an Intent ID and Proof of Intent Token are created.
- The agent uses a secure digital payment credential at checkout instead of the customer's actual credit card number.
- Amex said it has not yet seen an increase in disputes because of agentic commerce.
Why it matters. AI agents are starting to make payments for users, and the model's nondeterministic behavior can lead to errors. A standard that checks intent before payment could make agent-driven commerce safer for developers building these tools. Amex says the Intent ID can also help with fraud prevention and dispute resolution.
-
AI labs are buying data from dead startups to train models
- Startup shutdowns are rising, and a new group of firms helps founders sell their code and workspace data.
- SimpleClosure's Asset Hub has handled over 150 deals, and 47% of its deals now include code or data sales.
- Buyers include foundation AI labs, vendors that build simulated workspaces, and tech companies building their own models.
- In August, Google outbid Mercor to buy Spirit Airlines' data for $10 million, according to the article.
Why it matters. Real archives may help AI agents handle messy workplace tasks, but no company has published tests comparing them with synthetic data. Legal questions remain, including privacy and whether workers see any of the proceeds. Developers should watch how licensing and consent rules develop.
