Weekend roundup: Resurrected Clippy, ChatGPT Business and AI Oreo ads

Clippy much? Microsoft launches visualization of Copilot
If you ever use voice mode in Copilot, which Microsoft hopes to expand, you might see a new, expressive animation on your screen. That would be the newly announced «Mico.» Unlike the much maligned Clippy, Mico will use facial expressions that change as you talk. It’s only available in the US, and will work with an upcoming memory feature for Copilot to better respond to requests.
More at: Microsoft’s launch, The Verge and Ars Technica.

OpenAI announces ChatGPT Business
ChatGPT will now combine all the context of your businesses’ connected apps, like Slack, Sharepoint, Github and Google Drive. This makes it possible to ask pretty detailed questions about your business and have comprehensive answers delivered in one place — without the need to go searching through lots of different repositories. The feature is available tor Business, Enterprise and Education customers starting last Thursday.
More at: OpenAI’s launch page, The Verge and The Register.

Read on for more!

Snacks behemoth Mondelez betting big on AI savings
The Oreo maker and Cadbury owner has spent more than $40 million on an internal AI ads tool, looking to slash their production costs by 30—40%. It started last year as a collaboration between Publicis and Accenture, and it should be ready to produce short TV ads as early as the holidays of 2026 — and might feature at the 2027 Superbowl. They seem to have decided to not «yet» put human likeness in their AI commercials, after those that did were ridiculed.
More at Reuters.

$60 to turn off that «recording LED» on Meta’s Ray-Bans
As if it wasn’t creepy enough that anyone in the room might be silently recording you through their glasses, internet hacks have found a way to turn off the only protection there is — which is a LED lighting up when recording. The hack extraordinarily well executed, shows no external signs you are wearing modded glasses, and costs only 60 bucks. This LED feature is so important to Meta and its customers that the glasses won’t even work is you simply tape it over. But now the creeps found a way around that, which was just a question of time.
More at: Gizmodo and 404 Media.

Google signs up Anthropic for compute
Anthropic and Google have signed a deal worth «tens of billions of dollars» to secure a massive compute capacity for the AI Lab. Even if they are technically competing with Google, Google Cloud seems neutral and sits on a mind boggling compute capacity, that now will benefit Anthropic. The deal secures Anthropic access to up to a million Tensor processors and «well over» one gigawatt of capacity. The going rate for a 1 GW data center is around $50 billion, but the companies has not disclosed any dollar amount for the deal.
More at Anthropic’s announcement, CNBC, and Reuters.