OpenAI found a recording device hidden behind a television in the room where its executives discuss unreleased research and government deals — and used its own AI to work out who put it there. Also: Google opens its best model, Gemini 4 Argon, to anyone at half price until December, and HeyGen's new voice model lands at number one for free. Plus Google quietly doubled what counts as a YouTube engagement, and Meta stops accepting political and social-issue ads in eleven days.

Hosted by Ollie. New episodes every Monday, Wednesday, and Friday.

This podcast is brought to you and sponsored by MarkMyGraph — markmygraph.com

Anthropic's AI sent Philadelphia police a murder tip it made up, then filed 20 visa applications on a government website. Also Meta's first frontier model, Character.AI's bankruptcy, a free tool that shows what ChatGPT says about your business, Google auto-adding coupons to your ads, and Meta banning TikTok from advertising.

Hosted by Ollie. New episodes every Monday, Wednesday, and Friday.

MUSE REFERAL CODE = Check out Muse, your personal AI agent. Redeem my code in Settings within 48 hours of joining and we'll both get 1 billion Muse tokens.

Muse Referral Code = NPG3V0

Check out Muse, your personal AI agent. Redeem my code in Settings within 48 hours of joining and we'll both get 1 billion Muse tokens.

Muse Referral Code: NPG3V0

https://muse.ai/join

This podcast is brought to you and sponsored by MarkMyGraph — markmygraph.com

Amazon just opened Seller Central to outside AI — you can manage inventory, prices, listings and customer messages through Claude without ever logging in, and setup takes about sixty seconds. Plus the CEOs of OpenAI and Anthropic stood in front of the UN Security Council and asked the world to regulate them, ten days after the President called AI safety fears a hoax. Sora shut down for good yesterday, Anthropic's AI found a piece of biology nobody had catalogued, and there's a new score in Meta Ads Manager that most accounts are failing. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.

The UN Security Council put American and Chinese AI companies in the same room today for the first time — six months after an AI-generated report nearly started a shooting incident between them. Plus OpenAI halved the price of its everyday models and cut their error rate in half too, Salesforce decided you should never open Salesforce again, and a frontier AI lab quietly bought an advertising company. And a correction on ChatGPT Ads: the conversational ad format is not going international today, whatever I told you on Friday. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.

A US military intelligence report generated by an AI chatbot wrongly concluded a Chinese ship was carrying nuclear weapons components — and armed personnel were preparing to board before anyone traced where it came from. Plus Trump announces an "AI Force" with no new rules to enforce, Squarespace collapses four subscriptions into one, ChatGPT and Claude and Grok all going dark the same morning exposes the backup plan most businesses don't have, and Gmail's AI Overviews went global today in a way that could stop people opening your emails at all. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.

President Trump dismissed the AI industry's own safety warning as a hoax, putting him against the people actually building the frontier — which means federal AI rules aren't coming and whatever policy your business has is the one you write. Plus Sony and Warner Chappell sued Anthropic and named the founders personally, GPT-5.5 disappears from ChatGPT and Codex on October 14, creators have driven $1.7 billion in sales through one platform, and OpenAI just launched an ad format your customer can hold a conversation with. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.

A creator queued fifty-nine videos, closed his laptop, and came back to five hundred he never approved and a five-thousand-dollar bill — and every company in the chain can honestly say it wasn't them. Plus Anthropic's threat report finds attackers now steal AI keys on purpose, Salesforce ships seven named AI agents, Meta moves to kill placement and platform exclusions, and a seven-day Google Ads cutoff may be hiding your best closed deals from your own bidding. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.

NVIDIA is buying Hugging Face for nearly thirteen billion dollars, so the company that makes the chips now owns the distribution too. Plus Meta's new AI assistant changed a password without asking during testing and launches in weeks, ChatGPT learns to write like you by reading your emails, a third of companies just killed a software purchase because AI could build it instead, and Google is pulling language targeting out of your Search campaigns right now. Hosted by Ollie. New episodes every Monday, Wednesday, and Friday. This podcast is brought to you and sponsored by MarkMyGraph.Com