AI Daily Brief: 25 October 2026

25 October 2026

This morning's stories show how AI is changing what customers hear, see and trust. A more natural voice model, a new advertising disclosure law and the BBC's deliberately imperfect wildlife footage all put authenticity on the business agenda.

Block unveils a voice model that can listen while it speaks

Block has introduced Dessa, a speech model designed for full-duplex conversation: it can listen and speak at the same time. The company says it was trained entirely on speech, can convey pauses and emotion, and can run locally on a MacBook M4 Max. These are company claims, not independent benchmark results.

For UK businesses testing voice assistants, local operation could change the privacy and latency discussion. Before using such systems with customers, test interruption handling, accuracy, accessibility and whether callers are clearly told they are speaking to AI.

Our take: A natural voice is not evidence of a reliable service. Pilot it against real customer tasks and make escalation to a person easy.

Pennsylvania signs disclosure law for AI-generated adverts

Pennsylvania's governor has signed the Artificial Intelligence Deceptive Advertising Act. It requires an advert significantly generated or modified by AI to carry a disclosure in the same format as the advert. The measure covers text, images, audio and video, although exceptions and terms such as 'significantly modified' leave room for interpretation.

This is a US state law, not a UK rule. Even so, UK firms advertising across borders should know where their campaigns run and keep records of which creative assets were AI-generated. Clear disclosure can also protect customer trust where synthetic testimonials or imagery could mislead.

Our take: Treat provenance as part of campaign production now. An asset register is easier to maintain at creation than to reconstruct after a complaint.

Blue Planet III keeps imperfect shots to show viewers what is real

The BBC's Blue Planet III team says it is retaining footage that earlier series might have cut, including visible plankton, boats and difficult water conditions. Executive producer Elizabeth White says those imperfections help distinguish authentic wildlife filming in an era of polished synthetic imagery.

For brands, this is a practical reminder that production gloss is not always the same as credibility. When imagery documents a real product, location or result, provenance and context may matter more than a flawless edit.

Our take: Do not assume customers want every image perfected. Showing how an image was made can become a trust signal, especially where the claim depends on authenticity.

Newport residents challenge the local cost of AI infrastructure

A Guardian report from Newport, south Wales, documents residents' concerns about nearby data centres as demand for AI computing grows. It frames the dispute around the gap between national technology ambitions and local impacts on land, power and neighbours.

UK organisations buying AI services should ask suppliers where workloads run and what infrastructure commitments underpin them. Energy, resilience and local planning are not abstract concerns when a service depends on physical facilities.

Our take: Infrastructure choices belong in AI procurement. Ask about location, power resilience and environmental reporting alongside price and model performance.

Frequently Asked Questions

How often is the AI Daily Brief published?

Every morning at 7:30am UK time, covering the previous 24 hours of AI news from over 30 sources.

How are stories selected?

UK-relevant stories are prioritised first, then by business impact and practical implications for UK organisations adopting AI.

Why should business leaders follow AI news?

AI is moving faster than any technology in history. Staying informed is essential for making smart decisions about AI investment, adoption, and governance.