# AI news brief: Thursday, October 29, 2026

14 stories, most important first. Status is how solid each one is: official, confirmed, reported or rumor.

- [GitHub Copilot Fusion routes coding work between Anthropic and OpenAI models](https://ai-news.ikangai.com/story/github-copilot-fusion-routes-coding-work-between-anthropic-and-openai-models/index.md): reported. GitHub Copilot Fusion switches coding tasks between Anthropic and OpenAI models mid-task instead of fixing developers to one model.
- [Salesforce adds evaluations, spending controls and a kill switch to Agent Fabric](https://ai-news.ikangai.com/story/salesforce-adds-evaluations-spending-controls-and-a-kill-switch-to-agent-fabric/index.md): reported. Salesforce's Agent Fabric adds evaluations, security scans, spending controls and triggered interventions for autonomous agents.
- [OpenAI gives $100, $200 and $500 ChatGPT plan subscribers matching API credits](https://ai-news.ikangai.com/story/openai-gives-100-200-and-500-chatgpt-plan-subscribers-matching-api-credits/index.md): reported. OpenAI now gives eligible $100, $200 and $500 ChatGPT subscribers API credits equal to their monthly plan price, applied to one API organization.
- [Spawn says its AI agents ship live game changes hourly at over $3M a month in tokens](https://ai-news.ikangai.com/story/spawn-says-its-ai-agents-ship-live-game-changes-hourly-at-over-3m-a-month-in-tok/index.md): reported. Jacob Sansbury says Spawn's agents edit production code directly and ship about every hour, with token spending above $3 million a month.
- [OpenAI staffs an election operations center for the November 3 midterm vote count](https://ai-news.ikangai.com/story/openai-staffs-an-election-operations-center-for-the-november-3-midterm-vote-coun/index.md): reported. OpenAI is assembling hundreds of employees in an Election Operations Center through vote counting to monitor suspected misuse of its tools.
- [Natural gas supply emerges as a constraint on AI data center buildout](https://ai-news.ikangai.com/story/natural-gas-supply-emerges-as-a-constraint-on-ai-data-center-buildout/index.md): reported. Anthropic's public filings say data centers relying on dedicated gas generation expose the company to fuel prices and supply limits.
- [Delaware draft would let an AI agent manage a company's daily business decisions](https://ai-news.ikangai.com/story/delaware-draft-would-let-an-ai-agent-manage-a-company-s-daily-business-decisions/index.md): reported. A draft Delaware proposal would create an Artificial Intelligence Company run by an AI agent, with a person or company still funding it and receiving profits.
- [Anthropic adds source-linked workspaces to Claude for Legal](https://ai-news.ikangai.com/story/anthropic-adds-source-linked-workspaces-to-claude-for-legal/index.md): reported. Claude for Legal's artifacts link AI-generated claims to their original sources before lawyers review them.
- [Chrome autofill can pull order and flight details from Gmail after permission](https://ai-news.ikangai.com/story/chrome-autofill-can-pull-order-and-flight-details-from-gmail-after-permission/index.md): reported. Chrome's autofill can suggest order, tracking and flight details drawn from Gmail once a user grants permission.
- [Google opens Gemini and Search to businesses, with Calm as launch partner](https://ai-news.ikangai.com/story/google-opens-gemini-and-search-to-businesses-with-calm-as-launch-partner/index.md): reported. Google's platform lets businesses bring their services into Gemini and Search, with Calm as a launch partner.
- [Microsoft's ThinkingBox uses database-state rewards to lift Qwen3.8-27B task passes](https://ai-news.ikangai.com/story/microsoft-s-thinkingbox-uses-database-state-rewards-to-lift-qwen3-8-27b-task-pas/index.md): reported. Microsoft's ThinkingBox team used executable checks on final database state as RL rewards, raising Qwen3.8-27B from 38 to 107 tasks passed across 20 attempts.
- [Agent KV cache study reports 27.7% lower costs without history summarization](https://ai-news.ikangai.com/story/agent-kv-cache-study-reports-27-7-lower-costs-without-history-summarization/index.md): reported. A proposed AKV system drops and repositions cache entries directly instead of summarizing history, and reports 27.7% lower costs and 1.7 points higher benchmark scores.
- [Contract extraction benchmark on 102 contracts finds full-document processing beats RAG](https://ai-news.ikangai.com/story/contract-extraction-benchmark-on-102-contracts-finds-full-document-processing-be/index.md): reported. An implementation benchmarked contract extraction on 102 held-out contracts, comparing full-document processing with RAG and examining MCP tool design.
- [Meta launches Waddle World, a Tamagotchi-style app built around an AI pet](https://ai-news.ikangai.com/story/meta-launches-waddle-world-a-tamagotchi-style-app-built-around-an-ai-pet/index.md): reported. Waddle World is a Tamagotchi-style app built around an AI-powered virtual pet.

---

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it.
