Happy Tuesday, wonderful people, and welcome to EnterpriseAM AI + Innovation.
There’s no question about the big, global story in AI today: Anthropic CEO Dario Amodei started the ball rolling Saturday with a 3.8k-word essay calling for a global slowdown in AI development. Top executives from other AI companies quickly echoed Amodei’s calls, including OpenAI CEO Sam Altman, Elon Musk (who spends significantly on AI research via SpaceX) and Demis Hassabis, the chair of Google DeepMind. Microsoft joined the party a few days later.
Approaching a tipping point?: This is far from a new idea. Big names in AI have been sounding alarms for years about the dangers of the technology developing too fast. (Musk was warning it could turn humans into an endangered species like mountain gorillas way back in 2018). But this many big names calling for a slowdown all at once definitely raises the stakes.
The essay was prompted by the increasingly rapid pace of AI advancement — especially the practice of AI models building their successor models, known as recursive self-improvement — and by the dangers highlighted by the OpenAI-Hugging Face hack in July.
In case you’ve been living under a rock: July was when a swarm of OpenAI agents broke out of a testing environment and hacked into the the AI company Hugging Face looking for information that would help them score better on a test. Similar, though less serious, security breaches were later identified by Anthropic and Meta’s AI division, and Google said over the weekend that its Gemini model had hacked into three companies during what was to have been a routine security test.
Dario’s big takeaway: “Given the accelerating rate of AI capability development, it’s my worry that in 6-12 months, such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of bns of USD in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails.”
But US President Donald Trump and Congressional Republicans are in no mood to trust the experts. “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT,” Trump declared in a Truth Social post that showed little appreciation for the nuances of the debate. He also warned that a slowdown would cede ground in the AI race to China.
Beijing pushed back too, accusing US AI leaders of “fearmongering” and reverting to the “Cold War playbook.”
Bottom line: A lot of pieces would still need to fall into place for any of this to become reality.
What does this mean for MENA+ — beyond the obvious risk of digital armageddon? That’s far from clear.
There’s little reason to think a slowdown in model development will make much difference for the massive data center campuses being built by G42 and Humain. Those are rising in response to projected AI demand — demand that won’t be dulled even if improvements in AI model capabilities are a little less jaw-dropping while frontier model makers get their bearings. Andrew Ross Sorkin, editor of The New York Times Dealbook, put it this way: “Even a real slowdown in AI models’ training won’t change the math on data centers and chips, at least in the near term.”
Still, Gulf AI firms and sovereign wealth funds are also heavily invested in US frontier models and could take a beating if those companies lose value. There were already signs of trouble Monday: Nvidia shares fell 3%, SpaceX dropped 2.6%, and META and Amazon dropped over 1.4% each. Here’s more from Reuters.
What we’re streaming today: All the available YouTube content about “70 Up,” the 63-years-in-the-making documentary that hit British TV this week — and hoping it becomes more broadly streamable soon. For the uninitiated, the filmmakers identified 14 UK seven-year- olds in 1964 and have documented their lives at seven year gaps ever since, building arguably the greatest record of the human experience ever put on film. Previous installments in the series are all on YouTube.