# Weekly AI roundup: Anthropic safety, xAI Grok Voice 3.0, Google power

Digest AI · Business & Funding · published 2026-10-25T20:09:00Z

Canonical: https://digestai.news/story/weekly-ai-roundup-anthropic-safety-xai-grok-voice-3-0-google-power

## Summary

This weekly summary covers key developments from October 13 to 19 for five major AI companies. Anthropic announced safety measures for Claude regarding U.S. midterm elections, including granting independent evaluators at METR employee-level access to internal systems. The company also opened a Singapore office and proposed contributing 5% of its shares to a public interest fund, subject to shareholder approval.

OpenAI faced legal and transparency issues, with Reuters reporting that a former employee allegedly sent over 100 files related to unreleased products to a personal account after moving to Apple. Separately, reports highlighted disputes over the redaction of confidential information in independent AI evaluation reports. On the infrastructure side, American Electric Power plans a 702MW fuel cell facility in Wyoming for Google, with construction starting in 2026 and full operation expected by the end of 2028.

xAI released Grok Voice 3.0 on October 14, a speech-to-speech model with sub-one-second latency available via the Voice Agent API, while deprecating the previous version. Alibaba Cloud announced that support for some older Qwen 3.5 series models will end on January 19, 2027, urging users to migrate to successor models.

## Key points

- Anthropic granted METR employee-level access to internal systems for independent safety evaluations.
- xAI released Grok Voice 3.0 with sub-one-second latency, deprecating the previous version.
- Alibaba Cloud will end support for some Qwen 3.5 models on January 19, 2027.

## Why it matters

These updates highlight a shift toward transparency in AI safety and infrastructure sustainability. The deprecation of Qwen models requires immediate action from developers, while Anthropic's open access to evaluators sets a new standard for trust.

## Sources

1. [What happened this week with the top 5 generative AI companies? | A weekly summary of OpenAI, Google, Anthropic, xAI, and Qwen 10/13-10/19](https://note.com/zetton1983/n/n3193b8179e69?hl=en) (note.com, 2026-10-25)

Part of the developing story: [Anthropic Pentagon AI Clash Saga](https://digestai.news/thread/anthropic-warns-pentagon-dispute-could-cost-billions) (5 stories)

## Cite

Digest AI, "Weekly AI roundup: Anthropic safety, xAI Grok Voice 3.0, Google power", 25 October 2026, https://digestai.news/story/weekly-ai-roundup-anthropic-safety-xai-grok-voice-3-0-google-power

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/weekly-ai-roundup-anthropic-safety-xai-grok-voice-3-0-google-power.json