America’s plan to use AI as a political weapon

For those of us who are critical observers of the tech industry, an equivalent slogan now comes to mind: “The technological is political.” It captures a truth that Silicon Valley had been trying to ignore for at least three decades: the insistence that what it was building had nothing to do with politics – indeed was somehow above such mundane phenomena.

To those who thought they were changing the world, politics was just an occupation for lesser beings.

The last few years have rather taken the shine off that aloof disdain. First, the Silicon Valley crowd realised that the Biden administration thought they needed some – shock! Horror! – regulation. And second, the White House was very hostile to crypto. Worse still, the Biden administration was not overawed by tech geniuses.

Au contraire. Software engineer turned venture capitalist Marc Andreessen, one of Silicon Valley’s really big shots (and loudest mouths), was given the bum’s rush when he went to Washington. This convinced him, he said afterwards to Ross Douhat, that the tech industry was “no longer dealing with rational people. We’re no longer dealing with people we can deal with. And that’s the day we walked out and stood in the parking lot of the West Wing and took one look at each other, and we’re like: ‘Yep, we’re for Trump.’”

Which explains how all the big tech moguls wound up with front row seats at Donald Trump’s inauguration in January. Sidling up to Trump, though, was also shrewd because it meant that, suddenly, there was a chance that their corporate interests might become inextricably intertwined with those of the US government, to their mutual benefit.

That it was a good bet – for the industry, anyway – was amply confirmed last week when “America’s AI action plan” appeared. The overarching theme of the plan is to use the current US leadership in AI to build a global alliance of democracies using American technology while systematically denying adversaries access to critical capabilities.

It rests on three “pillars”: accelerating AI innovation in the US, building infrastructure for AI and making sure that the rest of the world signs up to the full US “AI technology stack” (hardware, models, software, applications, and governance and standards).

This third pillar is the really interesting one. It’s clearly aimed at countering Chinese influence in international AI governance bodies such as the UN, the Organisation for Economic Co-operation and Development, and the G7 and G20 groups – all bodies that China has been explicitly targeting with a proposal to “strengthen coordination to form a global AI governance framework that has broad consensus as soon as possible”.

At the core of the plan is an executive order from Trump – “Promoting the export of the American AI technology stack” – that is aimed at regions such as the EU and countries such as India that may have ideas of developing their own AI stacks.

“The United States,” says the action plan, “supports like-minded nations working together to encourage the development of AI in line with our shared values. But too many of these efforts have advocated for burdensome regulations, vague ‘codes of conduct’ that promote cultural agendas that do not align with American values, or have been influenced by Chinese companies attempting to shape standards for facial recognition and surveillance.”

Anyone tempted to take this as diplomatic blather should have a look at what’s going on in the US Congress, where a bi-partisan bill – the Chip Security Act – is on its way. This would require all AI processors (Nvidia H100s, for example) to be fitted with a location tracking device that will allow the US to monitor their movement across the globe. And perhaps also to add a facility to remotely disable the chip whenever it is being “misused”.

If you think there’s a whiff of paranoia here, join the club. Over in another part of the national security forest, a team led by Eric Schmidt, former chief executive of Google, has been mulling over a “superintelligence strategy” and has come up with MAIM – mutual assured AI malfunction – described as “a deterrence regime resembling nuclear mutual assured destruction (MAD) where any state’s aggressive bid for unilateral AI dominance is met with preventive sabotage by rivals”.

It envisages interventions ranging from covert cyber-attacks to potential kinetic strikes on datacentres, and “already describes the strategic picture AI superpowers find themselves in”.

So maybe we need an updated slogan: “The technological is now geopolitical.”

Maxwell’s silver hammer

What Epstein Was Afraid Of is Tina Brown’s brisk take on the scandal du jour and the late billionaire paedophile’s incarcerated ex-girlfriend Ghislaine.

All our yesterdays

A lovely essay by Ian Leslie on his Ruffian Substack is 27 Notes on Growing Old(er).

Wise words

Helen Lewis’s The Economics of Writing a Book offers useful wisdom for authors.

Photograph by Roy Rochlin/Getty Images