AI News Roundup — October 15, 2026
Three verified, sourced items from the past day (October 14-15) — a voice-model release, an election-integrity update, and a government AI partnership. Reported factually, no speculation.
xAI ships Grok Voice 3.0
xAI released Grok Voice 3.0, a speech-to-speech voice model available through its Voice Agent API at the same price as its predecessor. xAI says it ranks first on Artificial Analysis's Speech-to-Speech Index, responds in under one second, and scores 97% on Full Duplex Bench for handling turn-taking and interruptions — claims from xAI's own benchmarking that have not been independently verified by ModelScore.
Source: x.ai/news — "Introducing Grok Voice 3.0" (Oct 14, 2026)
Anthropic details election safeguards for the 2026 US midterms
Anthropic published an update on the steps it is taking ahead of the 2026 US midterm elections: Claude.ai will surface official, state-specific voting information sourced from Democracy Works and show official results on election night, and the company says it has added classifiers to catch multi-step misuse and banned accounts tied to fake news sites producing midterm coverage.
Source: anthropic.com/news — "Preparing for the 2026 US midterm elections" (Oct 14, 2026)
Google DeepMind and Singapore's GovTech build an on-device scam-detection model
Google DeepMind and Singapore's GovTech agency announced a privacy-preserving scam-detection model, fine-tuned from Google's lightweight Gemma model on local scam conversations so it can run directly on a phone and flag signs of manipulation in a chat. Because the model runs on-device, neither GovTech nor Google can access users' messages, per the announcement.
Source: deepmind.google — "On-device AI for scam detection, with Singapore's MDDI" (Oct 14, 2026)
Methodology: items are drawn from official vendor newsrooms where available, cross-checked against independent reporting. We link the primary source for every item and do not editorialize on capability claims we haven't verified ourselves. Items already covered in the prior day's roundup are not repeated here.