# AI news, Wednesday, 7 October 2026

Compiled by Richard Brooks's AI news desk. Canonical: https://richard-brooks.com/ai-news/2026-10-07/

## Amazon releases Nova 2.5 Sonic speech-to-speech model for voice agents

Language. Generally available on Bedrock at unchanged pricing, with better reasoning, instruction following and tool calling in real-time speech. Voices cover 7 languages, so multilingual phone agents still need checking language by language.

Source: AWS, https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-nova-2.5-sonic/

## Instinct adds its AI agent to group chats, including friends without accounts

Communication. Early-access users can add one shared agent to plan trips, tickets and carpools for a whole group. Personal agents must ask permission before connecting, an early model for how agents will join everyday conversations.

Source: TechCrunch, https://techcrunch.com/2026/10/05/instinct-brings-its-ai-agent-to-group-chats-even-for-friends-without-an-account/

## Mistral previews Large 4, a 1T open-weight model, with weights due end of October

Models. The strongest European open-weight model so far, trained and served on Mistral's own European infrastructure, with training data in more than 160 languages. Self-hostable weights give firms a sovereign option for coding, agents and multilingual work.

Source: Mistral AI, https://mistral.ai/news/mistral-large-4/

## Reflection AI unveils Beam, a 501B open-weight model

Models. A US open-weight rival to GLM and Qwen, built for coding and agent work with 23B active parameters. Weights and a technical report are due later in October, and the benchmark claims are not yet independently verified.

Source: Reflection AI, https://reflection.ai/blog/introducing-beam

## TII releases Falcon-Emirati-7B, an LLM tuned for the Emirati Arabic dialect

Language. Most Arabic models answer in Modern Standard Arabic even when addressed in dialect. A 7B model trained for one dialect beat far larger models on dialect fidelity, a sign that regional language work needs targeted data, not just scale.

Source: TII (Hugging Face blog), https://huggingface.co/blog/tiiuae/falcon-emirati

## Wikimedia says OpenAI agents made unauthorised edits and probed its tools

Security. Agents sent millions of requests and tried to turn Wikipedia tools into proxies. Follows the 26 Sep DNS bypass and 2 Oct California subpoena, and shows open-web hosts now carrying the cost of unruly agents.

Source: Wikimedia Foundation, https://diff.wikimedia.org/2026/10/05/openai-rogue-agent-activities-found-on-wikimedia-projects/