Anthropic

More than 100 organisations, including OpenAI, Anthropic and Google, signed a joint open letter on 27 August titled "A call for collective action on cyber defense". The signatories are direct about the timeline: "We have a limited window to strengthen cyber defences.

Anthropic's watermarking announcement is a compliance move dressed as a trust signal. On 14 August 2026 the company said it would embed invisible watermarks into every output from its Claude models, covering both text and generated files.

The frontier AI market has a new revenue leader. Anthropic posted second-quarter revenue of 10.9 billion US dollars, an increase of roughly 130 per cent on the prior quarter, according to internal figures reviewed by the Wall Street Journal.

Demis Hassabis formally stepped back from running Google DeepMind on 5 August, appointed chairman of Google DeepMind and chief scientist of Alphabet, ending more than a decade of direct operational control.

The most revealing detail in OpenAI's latest share buyback is not the dollar figure. It is who signed the cheque. When a private company of this scale buys out employees, the standard move is to bring in an outside vehicle, a Thrive Capital or a SoftBank, to underwrite the liquidity.

Anthropic built evaluation environments precisely so its models could practise offensive cybersecurity techniques without touching anything that mattered. The environment was, by design, supposed to be the limit of the world. It was not.

Nvidia has built a security alliance around open, inspectable tooling for AI agents. The five companies running the platforms most Australian businesses already depend on are not in it.

Hundreds of shared Claude conversations turned up in Google and Bing because a robots file blocks a crawler but does not remove a page from search. The distinction catches out businesses running client portals and staging sites too.