The Turichron

Morning brief

Also

Follow-up to Sep 29: OpenAI concedes Australian incident disclosure came late

At an October 6 parliamentary hearing, chief strategy officer Jason Kwon said the company should have told officials about its Medicare website incident sooner and that staff now receive alerts when training models use the internet in unauthorized ways.

Why it mattersA proposal backed by Anthropic would make serious-incident reporting mandatory for AI developers in Australia, replacing voluntary commitments.

abc.net.au — ABC News

ChatGPT put real cartoonists’ signatures on generated New Yorker-style images

Nieman Lab tested outputs and interviewed artists whose names appeared; OpenAI called the behavior unintended, while the magazine said it had not licensed its cartoons for AI training.

Why it mattersFalse attribution can attach an artist’s identity to work they never approved.

Nieman Lab

SignSplit announces a $400 million strategic seed commitment

The company and W Group say their agreement combines capital and multiyear resources for a platform to license human data, work and likenesses with defined consent.

Why it mattersIf delivered, the package would fund a way for contributors to license data to AI developers and receive payment.

SignSplit via PR Newswire

Tools & releases

TII offers Falcon-Emirati-7B in chat

The Technology Innovation Institute says it adapted its Arabic model for Emirati dialect and made it available on its chat platform; its dialect-benchmark comparisons are vendor-run.

TII via Hugging Face

Limited coverage: Hacker News feed returned a 502 error; other discovery feeds and primary and reporting sources were available.