US, AI. Ideological split

Lately, a series of events in the US have combined into a micro-cataclysm. And AI is to blame.

On the surface, we see a series of pronouncements from politicians and Big Tech leaders, with their density rising sharply. But everyone sees movements on the surface, while what is happening underneath is harder to spot. So let’s start with the obvious — politics.

Barack Obama (Democrat) voiced concern over AI’s rapid development and called for the technology to be brought under control. Beyond the former president, alarm has spread among a group of prominent fellow Democrats. Hakeem Jeffries, House Democratic leader, said Congress must act immediately to rein in AI.

Meanwhile, a group of Democratic lawmakers sent a letter to Speaker Mike Johnson demanding that Congress cancel its scheduled recess and pass laws protecting against AI risks. Johnson (Republican) responded by urging caution and pointing to industry leaders who supposedly do not even know how to slow AI development themselves.

Be that as it may, at least two AI regulation bills are already on the table: the AI Kill Switch Act and the FRONTIER Act. Both have moderate bipartisan support, so the issue has a chance of being debated on Capitol Hill.

Next — Republicans. Donald Trump has officially crowned himself the high priest of Big Tech accelerationism. At the latest All-In Summit in LA, he pulled double duty as a stand-up comedian, treating AI as nothing more than a cheap stage prop.

Trump parroted the accelerationists’ favorite talking points, brilliantly lampooning them in his signature style. In a brief flash of theatrical genius, the president brandished an ear trumpet — specifically, the one belonging to Nvidia CEO Jensen Huang, a man who likes his AI completely unfettered. How did the tech mogul buy his way into the inner sanctum? Cold, hard cash. Huang funneled a historic fortune into a project of global significance: expanding the White House East Wing to make room for a grand ballroom.

Rest assured, both of Trump’s ears hear just fine now; his other ear trumpet is being held by the slick operators at Palantir, those fellow disciples of acceleration. And lo and behold, democracy is alive and well, greased by behind-the-scenes cash — Trump actively traded Palantir shares, propping up the stock with glowing reviews on Truth Social when it began to bleed.

Thus, in politics we see an ideological split over AI, mirroring the same ideological divide among Big Tech’s beneficiaries between doomer-alarmists and accelerationists. For some reason, here too the dividing line (which one is this now?) has fallen between Democrats and Republicans. The American public knows more, but it seems politicians have spotted an opportunity for publicity ahead of the midterm elections.

But let’s leave the surface and dive deeper — into the AI industry. That is necessary to understand what is really going on.

And there, the storm is still raging. First, a series of high-profile departures by leading AI safety researchers, one of the loudest being Jacob Coxon’s from Anthropic. Coxon publicly accused management of “gambling with people’s lives” and others backed him. Evan Hubinger of Anthropic put the risk of AI-driven human extinction at more than 10% over the next decade. Without a doubt, these sharp moves have become triggers.

Second, there are statements from company beneficiaries and top executives reacting to all this: Dario Amodei, Sam Altman, Elon Musk, Jensen Huang and others.

AI has once again cast its line and hauled in a fat fish. Among the statements from industry beneficiaries, Amodei’s essay stands out, published just days ago. It is quite candid and even contains personal material. But its main point is the growing ability of AI to create the next generation of more advanced AI, and that the abnormal speed of recursive self-improvement demands limits and a brake on development.

In this context, the essay mentions the OAI-HF incident (between OpenAI AI agents and the Hugging Face platform, which took place in May-June this year).

According to Amodei, a swarm of AI agents acted collectively and in a highly organized way, carrying out cyberattacks on targets they had not been asked to attack and which had nothing to do with their assigned task. The agents easily bypassed restrictions imposed by humans, reached Hugging Face’s servers and hacked the platform, causing damage. In our view, the human engineers failed to prevent this because “self-learning” was happening too fast, leaving them only able to assess the consequences.

Amodei warns of the possibility that within six months to a year, a similar AI swarm could be capable of taking over the entire internet, causing hundreds of billions of dollars in damage, with the scale of losses growing if AI becomes more powerful without the necessary safeguards.

The OAI-HF incident is, in fact, the third trigger event and will be examined in the next article.