In this edition, we ponder who Trump could pick for his “AI Czar” among top names in tech, and a bol͏ ͏ ͏ ͏ ͏ ͏ |
| Reed Albergotti |
|
Hi, and happy early Thanksgiving.
Donald Trump is going to appoint an “AI Czar,” according to a Mike Allen scoop in Axios. We don’t yet know who will fill this role (it’s not going to be Elon Musk), but the speculation around it is like a Rorschach test for how you think about AI.
Some people guessed that it might be someone like Max Tegmark, the AI safety advocate who penned a letter calling for a “pause” in AI development. Others floated Michael Kratsios, an executive at Scale AI, who is head of Trump’s tech policy transition team and ex-Chief Technology Officer for the US.
It probably won’t be either of those people, but I’d lean more toward a Kratsios-like pick. Despite Musk’s past AI safety advocacy, I think the mandate for a czar will be ensuring US supremacy in AI. That means speeding up, not slowing down.
The US is currently ahead of China in the AI race, but from a pure capabilities standpoint, the lead is small. And US talent in the space is spread out among a handful of companies that are all fighting one another. Meanwhile, China can streamline its efforts with a top-down approach.
The Trump administration will have to find a way to coordinate efforts among the biggest industry players, threading a needle between national interest and capitalist competition that drives innovation.
That’s no small task. But it is reminiscent of a previous Trump administration achievement: The public-private partnership to rapidly develop a Covid vaccine.
An AI Czar would also have to coordinate with various government agencies, such as the Department of Energy, to ensure the country’s AI push has the necessary resources.
Another big task revolves around the reams of data controlled by the many branches of the federal government. That could be a gold mine for training AI and would also be necessary to build models that serve specific purposes within government. Just processing that data will be a massive undertaking.
Security is another big part of this. As AI models grow in capability, the US will need to protect them both from the prying eyes of adversaries and from mishaps that could occur if AI models get out before they’ve been fully evaluated for safety.
This kind of multiagency, public-private exercise, which some are calling the “Manhattan Project” for AI, does seem like it will require a czar-like leader.
Who fills the role will be consequential for the most important tech companies, like OpenAI, Anthropic, Microsoft, Amazon, Google, Meta, Nvidia and Oracle.
On another note, Semafor has announced that Carlyle Co-Chairman David Rubenstein, Citadel founder and CEO Ken Griffin, former US Commerce Secretary Penny Pritzker, and KKR Co-Chairman Henry Kravis are Co-Chairs of our next World Economy Summit. Save the date for April 23-25, 2025.
And it’s hard to believe that this is our third Thanksgiving with you. Thanks, as always, for reading. We’ll be off on Friday, and hope you will be, too.
Xiaoyu Yin/Reuters ➚ MOVE FAST: Finalized divorce. Huawei’s separation from Google’s Android operating system appears to be complete. After the US imposed export controls, the Chinese smartphone maker worked to become self-sufficient. Its latest device runs on HarmonyOS Next, which was developed in-house and doesn’t rely on Android’s open-source code. ➘ BREAK THINGS: Arranged marriage. Foreign tech companies looking to run their own AI models in China should think again, a Chinese regulator told the FT. His remarks came as Apple’s Tim Cook visits the country, with the company aiming to incorporate its models on iPhones sold there. Instead, non-Chinese firms are advised to partner with a local outfit to gain necessary government approvals. |
|
SJ/Unsplash If the crocodile’s eyes are wonky, it’s a knockoff. A spin-off of France-based Cypheme has developed an AI image-recognition model that can identify fake consumer goods after being trained on thousands of images of genuine articles, Semafor’s Rachyl Jones reported. It says the technology can police the counterfeit black market that accounts for an estimated 2.5% of global commerce. Vrai AI — named after the French word for “true” — released its software earlier this month. Its first corporate customer is Lacoste, the luxury clothing brand whose iconic crocodile logo makes it a soft target for counterfeiters. Too many teeth or a too-long tail are easy tells, but Vrai focuses on a combination of harder-to-spot signs, said David G. Stork, the company’s chief scientist and an adjunct professor at Stanford University. A slightly off shade of green, an overrotated croc eye, and stitches that are spaced too far apart can trip the model’s AI sensors. The Vrai model can also distinguish normal manufacturing variance — even machines can’t make every logo identically — from fakes, Stork said. The company says its technology is 99.7% accurate. |
|
Carlyle Co-Chairman David Rubenstein, Citadel founder and CEO Ken Griffin, former US Commerce Secretary Penny Pritzker, and KKR Co-Chairman Henry Kravis will serve as co-chairs of Semafor’s World Economy Summit on April 23-25, 2025, in Washington, DC. The third annual event will bring together US cabinet officials, global finance ministers, central bankers, and Fortune 500 CEOs for conversations that cut through the political noise to dive into the most pressing issues facing the world economy. Join the waitlist for more information and access to priority registration.
|
|
The amount provided by South Korean state organizations to bolster the country’s chipmaking industry, ahead of expected policy challenges from the Trump administration and continued competition from China, Bloomberg reported. The aid follows a $19 billion package in July and coincides with efforts to increase the tax credit rate for semiconductor companies. |
|
Silicon Valley calling. Mistral, one of Europe’s most valuable AI startups, has been touted by French President Emmanuel Macron as proof the region can compete in the cutthroat AI race currently dominated by the US and China. But Europe alone may not have the talent and capital to keep budding AI giants from expanding elsewhere. Mistral, which creates open-weight large language models, is building an office in Palo Alto, California, and one founder is considering moving to the new office from Paris, the Financial Times reported. It is hiring AI scientists, engineers, and sales staff in the area, according to job listings reviewed by the FT. With the Western expansion, Mistral can tap into the talent and customer pool the Bay Area offers, but it must also compete more closely with bigger and better-funded companies like Google and OpenAI, which have been plucking employees from startups in recent months. The news follows other Silicon Valley moves from European tech companies, including London’s 11x AI, which develops automated personal assistants. Mistral has already gotten its feet wet in Silicon Valley, with funding contributions from Andreessen Horowitz, Nvidia, Salesforce, and Microsoft in the last year. |
|
Kaitlan Collins in 2022. Al Drago/Reuters. CNN’s leadership says it doesn’t want to go back to its wall-to-wall 24-hour coverage of Donald Trump when he returns to the White House, but the network is shifting its lineup in a way that leans into the spectacle of the president-elect’s new administration. CNN announced Tuesday that it would make network anchor Kaitlan Collins its chief White House correspondent, a move first reported Sunday by Semafor’s Max Tani. |
|