While campaigners accuse Starmer of 'appeasing' big tech firms over online safety, a different kind of safety story has been quietly unfolding behind prison walls. Assaults on custody staff have dropped by 70% since the late 20th century, falling from 578 incidents in 1998 to just 173 today.

That contrast tells you something important about where safety improvements actually happen. Online safety generates headlines and political theatre. Prison safety doesn't. Yet one has measurably improved while the other remains mired in endless debate.

The numbers represent a fundamental shift in how Britain's most dangerous workplaces operate. In 1998, a prison officer faced roughly three times the risk of assault they do now. That's not a marginal improvement. That's a transformation.

What changed? Prison reform didn't make the evening news like social media regulation does. There were no parliamentary hearings about custody protocols. No campaigners demanding ministers take action on staff safety. The work happened quietly, through operational changes, better training, and reformed procedures.

The timing matters too. This 70% reduction occurred during decades when Britain's prison population surged, when overcrowding became chronic, when budget cuts hit the justice system hard. Staff didn't become safer because prisons became easier places to work. They became safer despite everything working against them.

Each of those 173 assaults still represents a real person going to work and getting hurt. Prison officers face violence that would be headline news in any other workplace. But the direction of travel matters. When politicians argue about regulating tech platforms to make the internet safer, they might look at what actually worked to make prisons safer.

The lesson isn't that online safety doesn't matter. It's that measurable safety improvements come from focused operational changes, not from political positioning. Prison staff didn't get safer because ministers gave speeches about custody reform. They got safer because someone redesigned how prisons actually function day to day.

As the debate over tech regulation continues, the prison data offers a different model. Real safety improvements happen when you measure the problem, change the systems, and track the results. Not when you hold press conferences about how seriously you take the issue.