Talk at the internal seminar for the ARIA 'Safeguarded AI' programme on propositional quantitative linear logic.