The description of the risk assessment tool launched in the period explicitly named attacks conducted with autonomous artificial intelligence as the problem it addresses.
That detail marks a transition. While a threat lives in research reports, it's a hypothesis; when it appears in a product's commercial description, it's because buyers recognise and budget for it.
The period assembled material explaining the shift: documented incidents with autonomous agents crossing trust boundaries, a new class of attack against coding assistants with a high success rate in testing, and a joint agency warning about offensive capability months away.
What changes in practice for defenders isn't the nature of the attack but the pace. Known techniques get executed with patience and speed no human team sustains.
The answer appearing in the period's products is always the same in essence: automate the defence to the same degree, because manual inventories and periodic reviews don't keep up with something scanning without pause.