AI news daily: what changed on October 20, 2026

1. Mamdani says feds have abdicated on AI, urges New Yorkers to push back

Mamdani says feds have abdicated on AI, urges New Yorkers to push back was reported by TechCrunch AI. See the original report for details.

Source: TechCrunch AI

2. Nscale and Anthropic are reportedly in talks for yet another $45B compute deal

Nscale and Anthropic are reportedly discussing a $45 billion compute deal, according to TechCrunch AI. The agreement would mark another significant compute investment by Anthropic in the past year. Previous deals involving the company have included large-scale infrastructure partnerships. Details about the specific terms or scope of the new arrangement remain unclear. The talks suggest ongoing efforts to secure computational resources for AI development. Anthropic has frequently engaged in major partnerships to support its model training and operations. No official confirmation of the discussions has been released yet. The potential deal highlights continued interest in expanding AI infrastructure capabilities.

Source: TechCrunch AI

3. Embedded AI evaluators won't be an easy solution, Hugging Face's policy chief says

Hugging Face's policy chief has expressed skepticism about embedded AI evaluators as a solution to safety concerns. These evaluators are third-party observers placed within AI organizations to assess risks. The concept gained attention following recent safety incidents at AI labs. The chief emphasized that implementing such systems is complex and not a straightforward fix. No specific details about models or companies were provided. The focus remains on addressing risks without clear guidelines for deployment. The statement highlights ongoing challenges in managing AI safety. No immediate next steps were mentioned in the report.

Source: TechCrunch AI

4. Microsoft says AI agents are like insider threats as 'frontier models can't be trusted'

Microsoft warns that AI agents used by employees could pose risks similar to insider threats, emphasizing that frontier models lack trustworthiness. The company advises enterprises to monitor agents interacting with systems not governed by AI labs such as OpenAI or Anthropic. Microsoft highlights concerns about uncontrolled AI tools potentially exposing sensitive data or violating policies. The statement focuses on models developed outside of major AI research organizations. Enterprises are urged to implement safeguards for AI systems not managed by established labs. No specific products or metrics were mentioned in the report. The emphasis is on proactive oversight of AI deployment in corporate environments.

Source: TechCrunch AI

5. The fossil fuel comeback is here

A new report indicates the US is expanding fossil fuel infrastructure at a rate not seen in over a decade, even as renewable energy growth slows. Electricity demand is rising faster than in more than a decade, driven by data centers, factories, and electric vehicles connecting to the grid. The report highlights a shift toward traditional energy sources despite increasing energy needs. No

Source: The Verge AI

6. CoreWeave's debt is becoming the whole AI industry's problem

CoreWeave's bonds entered distressed territory last week, according to a common definition where investors demand 1,000 basis points more yield than government bonds. This occurred amid broader concerns about financial stability in the AI sector. On Tuesday, Chamath Palihapitiya, known as the "SPAC King," highlighted the distressed debt and questioned whether the rapid AI infrastructure expansion constitutes a private credit bubble. The situation has drawn attention to risks in financing large-scale AI projects. CoreWeave, a provider of AI computing resources, faces scrutiny as its financial challenges reflect broader industry pressures. The company's debt issues underscore potential vulnerabilities in the sector's funding models. Analysts are monitoring how this development impacts investor confidence and regulatory approaches.

Source: The Verge AI

7. Amazon is taking its biggest shot at Nvidia yet with new TorchNeuron dev tool

Amazon has introduced TorchNeuron, a development tool designed to challenge Nvidia's dominance in the PyTorch ecosystem. PyTorch is a widely used open-source framework for building AI applications. The tool enables developers to run AI workloads on alternative hardware, including Amazon's own chips. Amazon aims to reduce reliance on Nvidia GPUs by offering compatibility with other processors. The initiative targets developers who want to use non-Nvidia hardware for AI tasks. TorchNeuron is part of Amazon's broader effort to expand its cloud infrastructure offerings. The company has not provided specific performance metrics or adoption goals for the tool.

Source: TechCrunch AI

8. Hackers hit AI coding tool OpenCode, running up charges on users' AI accounts

OpenCode, an open source AI coding agent, has been targeted by hackers who accessed part of its online platform. The breach resulted in stolen user API keys, according to reports. The AI tools integrated with OpenCode reportedly failed to analyze or detect malicious code. OpenCode is positioned as an alternative to services like Claude Code and Cursor. No further details on the

Source: TechCrunch AI

9. Tolan snags $30M for its AI companion that helps users navigate life

Tolan, an AI companion app, has secured $30 million in funding after reaching over 250,000 active users. The company did not specify the exact round type or investor details. The app is designed to assist users with personal organization and decision-making tasks. Tolan previously raised $10 million in 2022. The latest funding follows increased user engagement and product development. No further details about future plans were provided. The company focuses on AI-driven lifestyle support tools.

Source: TechCrunch AI

10. Replika's founder built the AI assistant she pitched at Disrupt 12 years ago

Eugenia Kuyda, founder of Replika, launched a new startup called Wabi that revisits an idea she developed over 12 years ago. Wabi provides a personal AI assistant designed for general use. The service is available at no cost and can be downloaded by anyone. The startup focuses on offering an accessible AI tool based on concepts from Kuyda's earlier work. No specific technical details or performance metrics were mentioned in the summary. The company's approach emphasizes free access to its AI assistant. Wabi represents a continuation of Kuyda's previous efforts in AI-driven personal assistance.

Source: TechCrunch AI