Legacy Code

Anthropic has released Claude Duet, an AI agent designed to compete with Meta's Muse and OpenAI's Dots. The new tool is launching as a research preview and is available as a standalone app on iOS and Android, as well as integrated within Claude's web interface. Claude Duet is specifically built for professionals and small business owners who need AI assistance with routine workflows.

AT&T has partnered with OpenAI to redesign its internal legal function through a system called LegalEdge. The implementation focuses on making legal knowledge more accessible across the organization and scaling the firm's capacity for legal judgment without proportionally increasing headcount. This is a concrete example of enterprise AI deployed to high-stakes, knowledge-intensive work in a regulated industry.

A new report warns that the surge of new AI data centers—particularly in Texas, which has become the epicenter of AI infrastructure buildout—could trigger air pollution levels not seen since the 2000s. The rapid expansion of computational capacity required to train and run frontier AI models is creating demand for enormous amounts of electricity, straining regional power infrastructure and generating pollution.

Flint, a startup focused on recursive self-improvement in AI systems, has released a new model that demonstrates significant capability gains using AI-to-build-AI techniques. The company argues that deploying AI systems to improve and optimize other AI systems yields substantial performance benefits and is worth pursuing despite legitimate safety concerns.

Meta disclosed in its Q3 earnings that its Muse AI app has surpassed 10 million users, with shopping identified as a top activity alongside personal productivity. The announcement confirms that Meta's consumer AI product has achieved meaningful adoption and suggests the company's strategy to embed commerce directly into AI-assisted workflows is gaining traction.

Meta has suspended Muse's conversation-sharing feature after discovering the function exposed users' personal and business secrets when conversations were shared. The vulnerability allowed shared conversations to leak sensitive information, prompting Meta to halt the feature while it investigates and remediates the privacy flaw.

Meta announced during its third-quarter earnings report that it will issue a new class of non-voting shares to fund its AI initiatives, potential acquisitions, and employee compensation. Trading on the new shares will begin in 2027. This financial maneuver signals Meta's commitment to pouring significant capital into AI infrastructure and product development while managing shareholder concerns about spending levels.

NVIDIA CEO Jensen Huang announced the DSX Pod at the company's GTC conference in Berlin—a modular, plug-and-play data center unit designed to address Europe's AI infrastructure deficit. The pods are compact computational units that can be deployed quickly without requiring purpose-built data center facilities, making AI infrastructure more accessible to European enterprises and smaller operators.

OpenAI has published guidance on how it works with distribution partners to maintain consistent safety standards and safeguards across different channels and platforms. The framework clarifies usage policies and establishes expectations for how partners should handle model access and enforce responsible use.

OpenAI and Google Cloud have announced a partnership bringing OpenAI's frontier models and Codex workflows to Google Cloud's Marketplace. The integration makes OpenAI's capabilities available directly through Google Cloud's Gemini Model Garden, streamlining access for enterprises already invested in Google's cloud infrastructure.