AI Watermarks, Royal Commissions

By the Orchestra editors · 4 min read

TL;DR

SA's AI Royal Commission catches the tech policy world off guard as states move ahead of Canberra
Anthropic's new Claude watermarks spark backlash from users worried about being outed
Google drops Pixel 11 lineup and Gemini hits 1 billion users
Massive supply-chain attack leaks terabytes of credentials via compromised AI package

AI · Worldwide

Anthropic's Claude watermarks anger users who'd rather not be caught

Anthropic rolled out watermarking for Claude-generated text and some users are furious, mainly because it makes it trivially easy to detect AI-written work in jobs and classrooms. The backlash is loud on social media but the company appears to be holding firm.

ChatGPT and Gemini both cross the billion-user mark

Google confirmed Gemini is its fastest product ever to reach a billion monthly users, with 63% using voice. OpenAI's ChatGPT quietly hit the same milestone a few weeks earlier, making it a two-horse race at enormous scale.

Compromised AI package leaks terabytes of credentials

A supply-chain attack targeting a widely used AI package scraped and exfiltrated credentials from roughly 2,500 users. A timely reminder that the rush to integrate AI libraries brings real security risk.

Australian Tech

SA's AI Royal Commission blindsides the policy crowd

Premier Peter Malinauskas announced a Royal Commission into AI, catching even seasoned tech policy watchers off guard. Combined with a reported OpenAI deal, it signals states are done waiting for Canberra to set the national AI agenda.

Experts flag AI agent liability after Australia's first automated hacking incident

Following what's been described as Australia's first reported automated hacking accident, legal experts say deployers of AI agents are on the hook for harm caused by their bots, even if unintentional. It sharpens the case for governance frameworks ahead of wider agent adoption.