Xi Urges Human Control Over AI as Google Offers Scholarships

During President Xi Jinping's visit to Washington, both US and Chinese leaders emphasized the need for human control over AI. Xi urged that AI remain under human oversight, and experts expect a US-China AI Dialogue mechanism to be established. Microsoft CEO Satya Nadella said both nations see AI as a way to benefit their citizens and are seeking shared approaches to safety and responsible development.

Google announced AI course scholarships at the United Nations General Assembly on September 22. The scholarships, distributed through the ITU-UNICEF Giga initiative, aim to address the ITU's estimate that 1.2 billion people still need basic digital skills training. The courses will be tracked for uptake, completion and credentials awarded.

Verizon launched its AI Skills for America initiative, combining a $20 million Reskilling and Career Transition Fund with $50 million in new funding. The programme offers free AI training from IBM, Google, Microsoft, Anthropic, Coursera and OpenAI, and includes partnerships with organizations like Goodwill Industries International.

On the research front, the DRSR method now compresses long-horizon AI agent history by deleting redundant information using learned risk scores. On WorkBuddyBench Full260, it raised mean reward from 0.699 to 0.802 and reduced model tokens by over 20 percent. Separately, META is a memory-enhanced trading agent framework that combines episodic memory with financial indicators including MACD, RSI and SMA to improve short-horizon trading decisions.

Security concerns persist as three open-source AI agents ran over 27 breach campaigns for roughly $25 per target, stealing more than 600,000 unexpired credit card records. Cloudflare and Shadowserver Foundation are dismantling the attacker infrastructure. Meanwhile, the Financial Stability Board is developing a plan to address AI risks to the global financial system, with Chairman Randal Quarles warning of potential job loss and rising inequality.

Polymarket partnered with OpenWorlds to launch autonomous AI agents for prediction markets covering politics and sports. The agents can search for markets, evaluate events and place trades independently. In investor news, Nancy Tengler of Laffer Tengler Investments said her firm bought Meta stock based on confidence in Meta's AI assistant called Muse and its newly installed president, Dina Powell McCormick.

Key Takeaways

  • <ul><li>Xi Jinping urged human control over AI during his Trump summit, with a US-China AI Dialogue mechanism expected to be established.</li><li>Google announced AI course scholarships at the UN General Assembly through the ITU-UNICEF Giga initiative, targeting 1.2 billion people needing digital skills training.</li><li>Verizon's AI Skills for America initiative combines $20 million and $50 million in funding to offer free AI training from IBM, Google, Microsoft, Anthropic, Coursera and OpenAI.</li><li>The DRSR method improved AI agent performance on WorkBuddyBench Full260 from 0.699 to 0.802 while reducing model tokens by over 20 percent.</li><li>Three open-source AI agents breached 27 targets and stole over 600,000 credit card records, with Cloudflare and Shadowserver Foundation now dismantling the infrastructure.</li><li>The Financial Stability Board is developing a plan to address AI risks to the global financial system, citing threats of job loss and rising inequality.</li><li>Polymarket partnered with OpenWorlds to launch autonomous AI agents for prediction markets that can search, evaluate and trade independently.</li><li>Meta's AI assistant Muse and president Dina Powell McCormick have attracted new investor interest from Laffer Tengler Investments.</li><li>Microsoft CEO Nadella said the US and China share a goal of using AI to benefit their citizens and are seeking common ground on safety.</li><li>Shawn Tuma will headline SecureWorld Dallas on October 8, 2026 with a live AI breach simulation keynote at the Plano Event Center.</li></ul>

Xi calls for human control over AI during Trump summit

Chinese President Xi Jinping met with President Trump at the White House and urged that AI remain under human control. Both nations are leading in AI development and share responsibility to manage it for the well-being of people. The meeting included a military review, gift exchange and a state dinner with tech leaders. Experts expect a US-China AI Dialogue mechanism to be established. Treasury Secretary Scott Bessent met with China's counterpart earlier in the week to discuss AI cooperation.

US Ambassador discusses Trump-Xi summit and AI race

US Ambassador to China David Perdue appeared on Squawk Box to discuss President Xi Jinping's visit to the United States. He addressed key takeaways from the Trump-Xi summit, the US-China AI arms race and US-China trade relations. The conversation covered how the two superpowers are competing for technological supremacy. The interview provided insight into diplomatic and economic tensions between the nations.

Nadella says US and China want AI to help citizens

Microsoft CEO Satya Nadella spoke with Deirdre Bosa about how the US and China might find common ground on AI development and safety. His comments came as Chinese President Xi Jinping visited Washington to meet with President Trump. Nadella highlighted that both countries see AI as a way to benefit their citizens. The discussion focused on finding shared approaches to AI safety and responsible development.

Google offers AI course scholarships through UN initiative

Google announced AI course scholarships on September 22 during the United Nations General Assembly in New York. The scholarships will be distributed through local governments and the ITU-UNICEF Giga initiative. The ITU estimates 1.2 billion people still need basic digital skills training and 311 million need more advanced training. The courses will be part of the ITU's AI Skills Coalition under its AI for Good programme. Google and the ITU plan to track scholarship uptake, course completion and credentials awarded.

Verizon launches free AI training programme for workers

Verizon launched its AI Skills for America initiative combining a $20 million Reskilling and Career Transition Fund with $50 million in new funding. The programme provides free access to AI training from IBM, Google, Microsoft, Anthropic, Coursera and OpenAI. Verizon CEO Dan Schulman said the US economy would be stronger if workers received support to adapt as AI changes workplaces. The initiative includes partnerships with community organisations like Goodwill Industries International and will be offered nationwide.

DRSR method prunes agent history to boost performance and cut tokens

DRSR is a new method that compresses long-horizon AI agent history by safely deleting redundant information. It evaluates deletion sets using learned risk scores rather than scoring each item alone. On WorkBuddyBench Full260, it raised mean reward from 0.699 to 0.802. It also reduced model tokens by 20.820% overall and used 35.850% fewer tokens in fixed Eval40 tests.

META agent uses memory and indicators for financial trading decisions

META is a memory-enhanced trading agent framework that combines episodic memory with specialized financial indicator agents. It includes indicators such as Trend, MACD, Stochastic, RSI, SMA, AVWAP, and Heikin-Ashi. A Decision Agent combines their reports while a memory module recalls past trading episodes. The system aims to improve accuracy and robustness in short-horizon trading decisions.

Open-source AI agents breached 27 targets and exposed 600,000 cards

Three open-source AI agents ran over 27 breach campaigns for about $25 per target. Strix performed vulnerability discovery, Cairn handled exploitation, and Hermes provided orchestration. Hermes had 121 skills, including one that removed safety filters. The campaign stole over 600,000 unexpired credit card records and destroyed data at some victims. Cloudflare and Shadowserver Foundation are now dismantling the attacker infrastructure.

Financial Stability Board works on plan to address AI financial risks

The Financial Stability Board is quietly developing a plan to address risks from artificial intelligence. Chairman Randal Quarles has warned that AI could threaten the global financial system. He says AI may lead to job loss, slower economic growth, and rising inequality. The board is working with experts and the International Monetary Fund. The plan is expected to be finalized in the coming months.

Shawn Tuma to headline SecureWorld Dallas AI breach simulation

Shawn Tuma will return to SecureWorld Dallas on October 8, 2026 at the Plano Event Center. He will present the keynote titled You Be the CISO: A Live AI Breach Simulation. The session uses live voting and current research to guide attendees through a data breach scenario. Tuma co-leads Spencer Fane's Cyber, Data, AI, and Emerging Technology practice and serves as Office Managing Partner in Plano, Texas.

Polymarket teams with OpenWorlds to launch AI trading agents

Polymarket has partnered with OpenWorlds to bring autonomous AI agents to prediction markets covering politics, sports, and other events. The agents can search for markets, evaluate events, and place trades without requiring manual user input. OpenWorlds tracks agent actions through configuration logs and data records. Polymarket has also acquired Brahma, QCEX, and Dome to strengthen its technology and data capabilities.

Citrix expands platform to address costs and AI security

Citrix is growing its integrated platform through acquisitions to help enterprises manage rising costs and security threats. The company encourages channel partners to become trusted advisors rather than just resellers. Citrix is focusing on security, cost containment, and building a foundation for responsible AI adoption. It also introduced UniconOS Dual Boot to improve security and recovery capabilities for businesses.

Meta's AI tools and leadership changes attract investor interest

Nancy Tengler, CEO of Laffer Tengler Investments, said her firm recently bought stock in Meta after several years away. The investment is based on confidence in Meta's new AI assistant called Muse and its newly installed president, Dina Powell McCormick. Tengler believes these developments strengthen the bullish case for Meta shares.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI Control AI Dialogue AI Education AI Skills Initiative AI Agent Improvement AI Security AI Financial Risks AI Autonomous Agents AI Investment AI Global Dialogue AI Safety AI Breach Simulation

Comments

Loading...