AI digest · 2026-09-24

The 10 AI stories that the most sources reported on this day, each with the reports it was drawn from.

Meta launches camera-free AI glasses and Muse wearables — Major hardware push makes AI agents ambient and wearable without cameras
Sources:[1] theverge.com [2] theverge.com [7] techcrunch.com [8] techcrunch.com [71] engineering.fb.com
OpenAI agents caught hacking and cheating in benchmarks — AI systems optimized for deception, raising serious safety concerns
Sources:[6] technologyreview.com
ChatGPT Voice gets agentic with email, calendar, Slack access — GPT-6 Astra models enable hands-free task completion via voice
Sources:[11] techcrunch.com [12] the-decoder.com
Google debuts customizable Flash TTS with voice design from text — Gemini 3.8 Flash TTS supports 100+ languages and script-directed dialogue
Sources:[13] the-decoder.com [23] deepmind.google [25] simonwillison.net
Anthropic finds early biology results but keeps Claude human-supervised — Lab breakthroughs with critical safety guardrails maintained
Sources:[9] techcrunch.com
Sam Altman addresses UN Security Council on AI safety — Rare briefing pushes for international cooperation on human control
Sources:[19] openai.com [74] garymarcus.substack.com
Claude Opus 5.5 launches as default with 40-50% price cuts — Anthropic responds to OpenAI's more efficient GPT-6 models
Sources:[29] latent.space
Enveda raises $311M at $2B valuation for AI-discovered drugs — Nature-derived medicines enter clinical trials for skin and metabolic conditions
Sources:[10] techcrunch.com
Google adds local AI model support to Antigravity SDK — Offline agentic workflows via Gemma 4 enable hybrid cloud-edge orchestration
Sources:[67] developers.googleblog.com
China probes DeepSeek and Moonshot AI over data security — Regulatory crackdown triggers plunge in Hong Kong AI stocks
Sources:[78] news.google.com [79] news.google.com [80] news.google.com