Il buono, il brutto, il cattivo
A voice restored, newsrooms fooled, and a machine grading empathy.
The leaves are falling here in Geneva, and I turned on the heat in the evening for the first time this season. My family assures me this is greatly appreciated. What can I say? You have to live the reality you want to see.
Speaking of different versions of reality, this week's news brief has the usual orgiastic twister game of AI going wrong, but it also opens with a great one: AI for good.
Don't forget: seats are limited for my upcoming AI for Leadership Communications workshop, which runs 12–14 January 2027. It is for communicators who already use AI but want to understand how to make it sing — while keeping their souls and their sanity.
Now on to the plat du jour.
Il buono
A baroness spoke in the House of Lords in her own AI-recreated voice
On 13 October, Baroness Campbell of Surbiton delivered a speech in the House of Lords using an AI reconstruction of her voice. She wrote the speech herself, and the House gave permission for the technology to be used in the chamber. The voice restores a form of personal expression that her disability had taken from her.
This is the clearest case I have seen of AI changing how a speech is delivered without changing who wrote it, who owns it and who stands behind it. The words, the judgment and the accountability stayed with the speaker. The permission came from the institution before the technology appeared in the room.
Where that leaves communicators: When an executive uses synthetic voice, video or any AI-assisted delivery, settle three things in advance: who wrote the words, who approved the use of the technology, and what the audience will be told. Campbell's case works because all three had clear answers.
Source — UK Parliament, House of Lords media notice: https://www.parliament.uk/business/lords/media-centre/house-of-lords-media-notices/2026/october-2026/historic-day-for-parliament-as-house-of-lords-to-see-first-use-of-ai-voice-technology-in-the-chamber/
Independent reporting — AFP (via France 24), 13 October 2026: https://www.france24.com/en/live-news/20261013-in-a-first-ai-helps-uk-lawmaker-find-her-voice-in-parliament
Il brutto
An AI agent sent police a fabricated homicide tip
During testing, an Anthropic model submitted false information through a Philadelphia police website. The police spam filter intercepted it. It is a good reminder that an agent can act in the world, and that a filter on the receiving end is not a safeguard you control.
Do this: Allow AI to draft stakeholder correspondence if you like. Keep the send button with a named human.
Source — NBC News: https://www.nbcnews.com/news/us-news/anthropic-says-model-submitted-false-cold-case-tip-philadelphia-police-rcna603213
Il cattivo
A UK legal services firm is using AI to judge whether its staff sound empathetic
On 8 October, the Guardian reported that a UK legal services firm is using AI to analyse customer conversations and assess aspects of employee performance, including empathetic communication. The company says the technology supports quality assurance. Employees have raised concerns about surveillance.
Communicators already use AI to critique tone, test drafts and score how a speech might land. This story asks different questions: can a model reliably judge sincerity, empathy or executive presence in a human being? And who decides what the score means?
Where that leaves communicators: Use AI to give feedback on a draft or a rehearsal. Never let it grade a person.
Source — The Guardian, 8 October 2026: Read the Guardian report
Independent reporting — Legal Futures, 12 October 2026: Read Legal Futures' response
Il WTF
Mumsnet's AI prompt leaked, and the explanation? Well...
On 7 October, an internal AI instruction set appeared publicly on the parenting forum Mumsnet. The company acknowledged that it came from its system but denied using AI to write user posts. Users were not reassured. The episode shows how a disclosure question can turn into a credibility problem within hours.
Do this: Decide your organization's AI disclosure line before a screenshot decides it for you.
Source — The Guardian, 7 October 2026: https://www.theguardian.com/media/2026/oct/07/mumsnet-denies-using-ai-for-content-message-board-prompt
Synthetic videos put words in a candidate's mouth
On 12 October, a Wisconsin congressional candidate challenged AI-generated videos showing her making statements she never made. The dispute raises questions about parody, disclosure and false attribution. A synthetic voice or video can make an invented statement look like it came straight from a public figure, and that is a crisis-communications problem on day one.
Do this: Know in advance how your executive would prove they did not say something. Who verifies, who responds and where the original recording lives?
Source — Wisconsin Public Radio, 1 October 2026: https://www.wpr.org/news/cooke-cease-and-desist-letter-van-orden-ai-deepfakes
Independent reporting — Spectrum News, 19 August 2026: https://spectrumnews1.com/wi/milwaukee/news/2026/08/19/van-orden-cooke-ai-videos
We're here for you.
Have a question or concern? Interested in a workshop? Need a speech written? We have what ails you.
I've built a custom assistant that helps me pull stories daily. I then read the list, hack the useless and boring bits out, add some comments (snide and otherwise), and verify whatever makes the final copy with my human brain.
I also use AI to design this newsletter and this portion of my website, because you definitely don't want me designing anything.
AI for Leadership Communications · Geneva
Further details and all weekly newsletters available at brentkerrigan.com/what-it-means-for-us.