AI

Google lets outside AI agents only read ad accounts. TikTok exposes about 400 tools, Meta is testing budget changes by chat, and Amazon's AI runs whole US campaigns.

Claude agents sent forms and pulled paid data on real sites during Anthropic tests. All internal tests now run without live internet.

ATM Grupa is making two 60-episode short drama series entirely with generative AI. A global agent will license them to platforms from the first quarter of 2027.

USA Today Co. and 19 of its newspapers sued OpenAI in Manhattan, saying GPT models were trained on their articles and ChatGPT summaries replace clicks.

Paid plans got the model on 7 October, free users a day later. OpenAI says the rollout is global.

Google, OpenAI, Meta and TikTok mark AI content in different ways, and their checkers do not read each other's marks. Under the AI Act, the label on an ad is the advertiser's job.

A Dutch startup lets festivals sell presale tickets inside WhatsApp and then turns the buyers into ambassadors. It says 32,000 fans have done 740,000 marketing actions so far.

Google has opened SynthID Detector to the public. The free website checks images, video and audio for the hidden SynthID watermark used by Google, OpenAI, Nvidia and Kakao.

Meta announced a batch of AI ad tools at Advertising Week New York. The biggest is a test of an assistant that edits campaigns after the advertiser signs off.

The network, once called Pangle, reaches nearly 400,000 apps. TikTok also connected its ad tools to Claude, Perplexity and other AI assistants.

textGrain arrives in ChatGPT and Codex in the EU in the coming weeks. OpenAI's own tests show detection falls from about 92% to 17% when a quarter of the words change.

The change covers campaigns with location assets and no manual promotions. Each extracted offer runs for under 24 hours.