# AI News Oct 16, 2026: Meta Muse Spark Open, ChatGPT Reads Chrome Tabs

> Meta launches an open model running on Macs, ChatGPT reads Chrome tabs, Google debuts Universal Cart in the US, and the FTC probes OpenAI and Anthropic. Business impacts summarized.

- Author: Nguyễn Ngọc Độ
- Published: 2026-10-16
- URL: https://nguyendo.id.vn/en/blog/ai-news-oct-16-2026-meta-muse-spark-open-chatgpt-reads-chrome-tabs

## Key takeaways

- Meta announced Muse Spark Open, an open-weight model reportedly capable of running locally on Apple silicon through a partnership with Apple.
- ChatGPT's Chrome extension can now view open tabs when users tag Chrome in the ChatGPT web or desktop app.
- Google rolls out its Universal Cart shopping tool in the US across Google Search and the Gemini app.
- Regulatory pressure on AI intensifies: The FTC is reportedly probing OpenAI and Anthropic, while Anthropic grants METR employee-level access.
- Vietnamese startup Drylab AI raises $3M in seed funding to build AI tools for scientists.

Today’s top stories center on AI moving deeper into our devices and daily shopping habits: Meta launched an open model that runs on Macs, ChatGPT can now read browser tabs, and Google introduced AI shopping carts into Search and Gemini. Alongside these developments, governance and accountability for major AI players are heating up. For enterprises, now is the time to weigh both implementation opportunities and data risks.

## Meta Releases Muse Spark Open, an Open Model Running on Mac

Meta introduced Muse Spark Open, claiming it to be the most powerful US open-weight model to date. According to Meta, it is also the first model in its class capable of running on desktops using two Apple M5 Ultra Mac Studio machines powered by the latest MLX framework. The model was co-developed with Apple to run locally on Apple silicon.

**Why it matters:** Local open models unlock the ability to process proprietary data in-house without sending it to the cloud. However, the hardware requirements mentioned are quite steep, so small businesses must carefully calculate hardware costs before setting expectations. Read more on [TechCrunch](https://techcrunch.com/2026/10/15/meta-launches-an-open-weight-model-that-runs-on-macs/) and [The Verge](https://www.theverge.com/news/1012293/meta-apple-muse-spark-open-ai-model-mlx).

## ChatGPT Can Now View Your Open Chrome Tabs

OpenAI upgraded the ChatGPT Chrome extension. Once the extension is installed and Chrome is tagged in the ChatGPT web or desktop app, ChatGPT can see your currently open tabs. According to The Verge, the ChatGPT Atlas browser is no longer in development, shifting the current strategy toward augmenting existing web browsers.

**Why it matters:** AI assistants are evolving from standalone chat boxes to working directly within an employee's workflow context, raising critical questions regarding access permissions and sensitive data across tabs. Read more on [The Verge](https://www.theverge.com/ai-artificial-intelligence/1012510/openai-chatgpt-google-chrome-tabs).

## Google Launches Universal Cart in the US

Google rolled out its Universal Cart shopping tool to US users, available on Google Search and the Gemini app. The brief reports did not share further technical details on how it functions under the hood.

**Why it matters:** As the purchasing journey shifts directly into Search and AI assistants, brands need to ensure product data, pricing, and inventory are accurately represented wherever customers search. Read more on [TechCrunch](https://techcrunch.com/2026/10/15/googles-universal-cart-shopping-tool-launches-in-the-us/).

## FTC Reportedly Probes OpenAI and Anthropic; Anthropic Opens Doors to METR

According to a report, the FTC is investigating whether security incidents at OpenAI and Anthropic harmed American consumers. Penalties could be comparable to those levied against companies that delay reporting data breaches. On the same day, Anthropic granted METR researchers employee-level access, allowing METR to independently publish evaluations of the company’s risks, incidents, and practices. Anthropic has waived the right to block these publications.

**Why it matters:** AI providers face mounting pressure regarding transparency and security accountability. When choosing AI partners, enterprises should explicitly ask about incident response and disclosure protocols. Read more on [Ars Technica](https://arstechnica.com/tech-policy/2026/10/ftc-probes-if-openai-anthropic-security-incidents-harmed-americans-report-says/) and [TechCrunch](https://techcrunch.com/2026/10/15/anthropic-is-letting-metr-researchers-look-under-the-hood-with-employee-level-access/).

## Amazon Eyes Ambient AI Devices, Launches Two New Bee Devices

At TechCrunch Disrupt, Amazon devices chief Panos Panay outlined a vision of ambient AI devices operating in the background, proactively anticipating needs rather than awaiting commands. Concurrently, Amazon-owned Bee launched two new personal AI devices: an updated wearable and a credit-card-sized voice recorder that attaches to phones. They can summarize conversations, generate to-do lists, and integrate with Alexa and Amazon shopping, starting at $74.99.

**Why it matters:** If dedicated AI hardware gradually replaces launching apps, how customers access services could fundamentally change, requiring businesses to consider making their services invokable via assistants. Read more on [TechCrunch (Panay)](https://techcrunch.com/2026/10/15/amazon-devices-chief-panos-panay-sees-a-future-where-ai-devices-replace-the-need-to-use-apps/) and [TechCrunch (Bee)](https://techcrunch.com/2026/10/15/amazons-bee-launches-two-new-personal-ai-devices-starting-at-75/).

## Figma Defines How AI Agents Work on Its Platform

Figma announced new features aimed at defining how AI agents operate across its platform. The company also pushed its code layers feature into open beta.

**Why it matters:** Design tools are becoming environments where agents actively participate in the product workflow; product teams should follow closely to streamline the design-to-code pipeline. Read more on [TechCrunch](https://techcrunch.com/2026/10/14/figma-announces-new-features-to-define-how-agents-work-on-its-platform/).

## Bluesky Lets Users Build AI Games and Websites Directly on Social Media

Bluesky expanded Attie, its experimental AI-powered platform, toward "social vibe coding." Users can build games, quizzes, polls, dashboards, or websites, and share them as standalone web apps directly with their social audience. Attie was originally launched to allow users to build custom feeds.

**Why it matters:** Building mini-apps using natural language drastically lowers barriers to entry, enabling brands to experiment with interactive micro-apps for customer engagement. Read more on [TechCrunch](https://techcrunch.com/2026/10/15/bluesky-expands-its-ai-powered-attie-app-to-let-anyone-build-websites-games-and-more/) and [The Verge](https://www.theverge.com/tech/1012374/bluesky-ai-attie-small-social-games-apps).

## Vietnamese-Founded Drylab AI Raises $3M

Drylab AI, a startup founded by three Vietnamese entrepreneurs, raised $3 million in a seed round led by Vertex Ventures Southeast Asia & India, with participation from PatSnap. The startup focuses on AI tailored for scientists.

**Why it matters:** This signals that regional venture capital is eager to back Vietnamese-founded AI startups that target clear, deep-vertical use cases. Read more on [VnExpress](https://vnexpress.net/startup-viet-goi-von-3-trieu-usd-cho-ai-ho-tro-nha-khoa-hoc-5132272.html).

## Strategic Takeaways for Enterprises

- **Audit AI tool permissions:** As assistants gain access to browser tabs, organizations must clearly define acceptable use policies regarding customer data, billing, and contracts.
- **Consider local models for sensitive workloads:** Running open models locally is worth evaluating, but factor in hardware procurement and operational talent before committing.
- **Standardize product and pricing data:** With commerce occurring natively in Search and AI assistants, accurate product specs, pricing, and inventory feeds are essential for visibility.
- **Incorporate security criteria into AI vendor selection:** Inquire about incident handling, breach disclosure practices, and transparency policies as scrutiny on AI providers escalates.
- **Run small pilots before scaling up:** For nascent tools like design agents or AI mini-apps, conduct controlled pilot projects with clear metrics rather than overhauling entire processes at once.

## Sources

- [Meta launches an open-weight AI model that runs on Macs](https://techcrunch.com/2026/10/15/meta-launches-an-open-weight-model-that-runs-on-macs/)
- [Meta and Apple team up to make Muse AI models run better on Macs](https://www.theverge.com/news/1012293/meta-apple-muse-spark-open-ai-model-mlx)
- [ChatGPT can peek at your Chrome tabs now](https://www.theverge.com/ai-artificial-intelligence/1012510/openai-chatgpt-google-chrome-tabs)
- [Google’s Universal Cart shopping tool launches in the US](https://techcrunch.com/2026/10/15/googles-universal-cart-shopping-tool-launches-in-the-us/)
- [Report: FTC probes Anthropic, OpenAI after Trump dismissed AI fears as “hoax”](https://arstechnica.com/tech-policy/2026/10/ftc-probes-if-openai-anthropic-security-incidents-harmed-americans-report-says/)
- [Anthropic is letting METR researchers look under the hood with employee-level access](https://techcrunch.com/2026/10/15/anthropic-is-letting-metr-researchers-look-under-the-hood-with-employee-level-access/)
- [Amazon devices chief Panos Panay sees a future where AI devices replace the need to use apps](https://techcrunch.com/2026/10/15/amazon-devices-chief-panos-panay-sees-a-future-where-ai-devices-replace-the-need-to-use-apps/)
- [Amazon’s Bee launches two new personal AI devices, starting at $75](https://techcrunch.com/2026/10/15/amazons-bee-launches-two-new-personal-ai-devices-starting-at-75/)
- [Figma announces new features to define how agents work on its platform](https://techcrunch.com/2026/10/14/figma-announces-new-features-to-define-how-agents-work-on-its-platform/)
- [Bluesky expands its AI-powered Attie app to let anyone build websites, games, and more](https://techcrunch.com/2026/10/15/bluesky-expands-its-ai-powered-attie-app-to-let-anyone-build-websites-games-and-more/)
- [Bluesky’s AI tool now lets you make games and apps to share on social](https://www.theverge.com/tech/1012374/bluesky-ai-attie-small-social-games-apps)
- [Startup Việt gọi vốn 3 triệu USD cho AI hỗ trợ nhà khoa học](https://vnexpress.net/startup-viet-goi-von-3-trieu-usd-cho-ai-ho-tro-nha-khoa-hoc-5132272.html)

## FAQ

### What is Meta's Muse Spark Open?

Muse Spark Open is Meta's new open-weight AI model, claimed to be the most powerful US open-weight model to date. Meta states it can run locally on Apple silicon, specifically on two Apple M5 Ultra Mac Studio systems using the latest MLX framework.

### How does ChatGPT view Chrome tabs?

After installing the ChatGPT Chrome extension and tagging Chrome in ChatGPT's web or desktop app, ChatGPT can see your active tabs. Businesses should carefully assess sensitive data in open tabs before enabling this feature.

### Is Google Universal Cart available in Vietnam?

According to reports, the tool has only launched in the US across Google Search and the Gemini app. Sources have not mentioned rollout plans for other markets yet.

### Why is the FTC investigation into AI labs important for enterprise AI adopters?

Reports indicate the FTC is reviewing security incidents at OpenAI and Anthropic, with potential penalties matching those for delayed data breach disclosures. This underscores the need for enterprises to scrutinize the reliability and security commitments of their AI vendors.