These are the "duties" Ofcom is demanding of site owners and operators:

  - to conduct a suitable and sufficient illegal content risk assessment;
  - to use proportionate measures to prevent individuals encountering priority illegal content;
  - to use proportionate systems and processes to minimise the length of time priority illegal content is present;
  - to swiftly take down illegal content when it becomes aware of it;
  - to specify in its terms of service how individuals are to be protected from illegal content; and relating to content reporting and complaints procedures in relation to illegal content.

Reads to me like they're legally prescribing that all web content is to be automatically fed into AI models in order to assess the content's appropriateness and "risk", and then "swiftly" censor the content if it's deemed too risky or illegal.

That's the only method that will scale, can be done "swiftly" and won't have the government kicking down your door if someone posts something illegal on your platform while you're sleeping or on vacation.