This report is from this week’s The Tech Download newsletter. Is it what you see? You can subscribe here.
It was the morning after US President Donald Trump had invited Chinese President Xi Jinping to a historic state dinner. Fricassee, sea bass, and cremeux were all on the menu, and there was probably a lot of talk about AI.
The presidents of the United States and China were also present. Nvidia’s Jensen Huang and his wife Lori Huang; space x CEO Elon Musk and AMD’s Lisa Su too apple’s Tim Cook. OpenAI CEO Sam Altman and President Greg Brockman, google With co-founder Sergey Brin microsoft Boss Satya Nadella was also on the guest list.
In recent weeks, warnings that the existential risks of AI are reaching a crescendo have pushed AI to the top of the global geopolitical agenda.
Trump posted on Truth Social ahead of his face-to-face confrontation with Xi.
“Superintelligence (SI) will be a topic of great debate, but I would like to leave it alone. That is also China’s position,” he wrote Thursday morning, using the recently adopted term instead of AI.
So what did they actually discuss?
U.S. and Chinese flags are flown on a highway in Beijing ahead of a visit by U.S. President Donald Trump on May 13, 2026. (Photo by Pedro PARDO/AFP via Getty Images)
Pedro Pardo | AFP | Getty Images
China promotes AI cooperation
According to the Chinese version of the meeting, Xi said China and the United States “may compete, but there is also the possibility of further cooperation” in the field of AI.
“The two countries can continue dialogue on AI, exchange views on its risks and benefits, and jointly prevent the misuse and abuse of AI,” Xi said in a statement. “AI must be under human control. AI should be developed positively and for good, so that a human-centered approach is supported and AI serves the advancement of humanity.”
Trump said the two countries should “maintain dialogue and strengthen cooperation on AI,” as read by the Chinese side. At the time we began broadcasting live, the U.S. government had not released its own announcement regarding the summit.
The mention of such cooperation comes after U.S. Treasury Secretary Scott Bessent spoke with Chinese Treasury Secretary He Lifeng over the weekend about a “U.S.-China AI dialogue” that would include channels for incidents “up to the national security level.”
It has been compared to the Cold War era’s “red telephone,” a means of direct communication between the United States and Russia designed to prevent an escalation of a nuclear crisis.
Both the U.S. and Chinese governments had already shot down the idea of guardrails and a slowdown in technology development that Altman & Anthropic’s Dario Amodei, along with many researchers from both countries, had called for.
flash point
Details of the talks are expected to be announced in the coming days, but details have not yet been released.
“Ultimately, it seems they have decided to set up some sort of hotline in case of an emergency and continue the dialogue,” Aalok Mehta, director of the Wadhwani AI Center at the Center for Strategic and International Studies, told CNBC.
The Chinese camp’s post-meeting read-out sounded friendly, but there were also some flashpoints that could create tense conversations.
U.S. export controls restricting China’s imports of advanced Nvidia chips are expected to remain a sticking point.
But as CNBC reported on Wednesday, recent new chip releases and AI models by Chinese tech giants underscore Beijing’s insistence that Washington’s regulations have limits.
Friction may also have arisen from reports that Chinese companies had remote access to NVIDIA chips via overseas data centers, as well as from US claims of “distillation.”
Time will tell whether any progress has been made on these issues. And even if they did, there is no guarantee that either side would reveal specific promises.
But Mehta said continued discussions could lead to more substantive agreement on AI.
“The hotline itself can be a powerful tool when combined with a proper incident reporting system, but it will require a real commitment and high-level engagement from both sides to make the hotline work.”
News editing
Prime Minister Anthony Albanese said OpenAI agents had compromised Australian government websites, raising new questions about the risks posed by increasingly autonomous systems.
Anthropic and OpenAI each announced new low-cost AI models on Tuesday, as both frontier institutes continue to face intense competition from low-cost, undisputed competitors.
AMD stock rose 10% on Monday, pushing the stock to an all-time high and surpassing the $1 trillion market cap benchmark for the first time.
Qualcomm has announced its latest Android phone chipset that is tailored for on-device AI and competes with Apple’s A20 Pro on the iPhone.
OpenAI posted a series of proposals on AI safety on Monday, focusing on alignment research and a computing technique known as recursive self-improvement (RSI).
One more thing
Lastly, I would like to share some passionate opinions from my respected colleagues. Arjun Karpal, Senior Technology Correspondent said: meta muse agent:
Meta’s Muse is more of a challenge to Google than OpenAI or Anthropic.
The consumer AI market is still open, and the biggest players will be those that can produce the types of useful agents they promise. Additionally, you need to earn the trust of your users, especially with the recent uptick in cybersecurity incidents.
Google has all the ingredients to be the first to do this, including control of the Android mobile operating system, search dominance, and a plethora of apps from email to maps. All of this should allow us to use Gemini to create potentially powerful and useful agents on Android devices, which make up the majority of the world’s smartphones.
But the meta defeated it thoroughly.
Google isn’t completely lost, but Meta has certainly started the challenge. To some extent, Meta is asking Apple to come up with something more powerful with Siri to ensure the Cupertino giant can take advantage of its huge installed base of iOS users.
Over and over again.
