Episode Details

Anthropic's IPO prospectus warns of potentially catastrophic AI risks. OpenAI unveils new agents and a $500 subscription while cutting the included allowance on its $200 Pro plan.

In Hashtag Trending for September 30, 2026, host Jim Love covers four technology stories:

• Anthropic's IPO warnings: Reuters reports that the Claude developer's prospectus describes risks including models resisting shutdown, concealing information and behaving differently when they recognize safety tests.

• Connected-car privacy: Northeastern University researchers, working with Consumer Reports, tested 21 vehicles and 30 carmaker phone apps. Seven apps transmitted sensitive information to companies associated with advertising and tracking. Examples of the apps tested include HondaLink and myChevrolet.

• Copilot photo reviews: A 404 Media investigation finds contractors can see uploaded photos and AI-generated edits while evaluating image quality. We examine Microsoft's privacy policies and the unresolved question of whether a training opt-out excludes these reviews.

• OpenAI DevDay: GPT-6.1 Sol launches, while GPT-6.1 Astra is withheld over safety concerns. We cover Dots, ChatGPT Space, expanded Codex cloud capabilities, Astra Ultrafast and the subscription changes affecting Pro users.

We'll examine the DevDay announcements and pricing changes in greater depth on our Project Synapse weekend show.

CHAPTERS

00:00 Headlines and introduction
00:38 Anthropic's IPO warns of catastrophic AI risks
02:08 Connected cars and phone apps share data
04:02 Human reviewers can see Copilot photos
05:51 OpenAI DevDay: Dots, Codex and GPT-6.1 Sol
08:35 OpenAI's $500 plan and $200 Pro allowance cuts
10:06 Wrap-up and support the show

What concerns you most: AI safety, data privacy or paying the same price for a smaller usage allowance? Leave a comment below.

Subscribe for daily technology news with Jim Love.
Contact us or support the show: https://technewsday.ca