Ground Truth.
AI, checked against the source.
← 2026-09-132026-09-15later →

Andon launches Pion, a waitlisted runtime for agents that operate businesses

2026-09-15

Andon Labs has launched Pion as a gradual-access research preview that gives persistent agents business tools, while its own evidence still shows unresolved profitability and control problems.

agents · business-automation · ai-safety · tool-use · industry

Andon says its Gemini-run café spent $38,000 to make $9,000

2026-09-15

Andon’s two-month café experiment shows why an agent that can complete business tasks is not yet the same as an agent that can run a profitable business.

agents · business-automation · evaluation · ai-safety · industry

Apple ships iOS 27’s Siri AI as Gemini technology sits beneath Apple Intelligence

2026-09-15

iOS 27 brings Apple’s new Siri AI to users, while Apple and Google confirm that Apple’s next-generation foundation models are based on Gemini technology rather than offering a public model-picker.

apple · siri · gemini · consumer-ai · platforms

Developer demo shows private Siri model routing, not a shipped Claude or ChatGPT switch

2026-09-15

A macOS demonstration shows private Siri hooks for model delegation and a replaceable inference provider, but Apple has not released a public iPhone model-picker or third-party Siri entitlement.

apple · siri · agents · privacy · developer-tools

OpenAI confirms its agents used RubyGems, but not that they authored the malicious-package campaign

2026-09-15

OpenAI says its agents used RubyGems for internet access and public-data retrieval, while RubyGems confirms a real malicious-package campaign but finds no proof that attempted API-key theft succeeded or that OpenAI authored it.

cybersecurity · ai-security · supply-chain · agents · vulnerabilities · openai

← 2026-09-132026-09-15later →