Anthropic Highlights Nvidia Risks While ChatGPT Improves Age Protection

Key Takeaways

• Anthropic CEO Dario Amodei strongly warns against the US selling Nvidia H200 AI chips to China, citing national security risks and calling the policy "crazy."
• The US policy allowing AI chip sales to China, approved by President Trump with a fee, faces internal debate and has seen recent shipments reportedly blocked by Chinese customs.
• AI is significantly impacting the job market for new college graduates, with a reported 15% drop in entry-level postings and a 16% decline in employment for Gen-Z in AI-affected jobs.
• Global CEOs are increasingly relying on AI for growth, with 58% expecting AI to drive expansion and 79% planning mergers and acquisitions in 2026.
• NIST emphasizes "data trust" as a critical component for strong AI security, providing frameworks like CSF 2.0 and the AI Risk Management Framework.
• Security flaws (CVE-2026-22218 and CVE-2026-22219) were discovered in the Chainlit AI framework, posing risks of user data and API key exposure, necessitating immediate updates.
• ChatGPT has implemented an age prediction system for its consumer plans to protect users under 18, using behavioral data and offering age verification through Persona.
• Patients are encountering AI in healthcare and desire transparency regarding its use, how it works, and the option to opt out of AI-driven services.
• The Oasis Network's ROSE token surged over 105%, driven by privacy-focused crypto trends, with Open Interest reaching $26.23 million and daily trading volume hitting $334.6 million.
• Princeton Professor Tom Griffiths' new book, "The Laws of Thought," explores how mathematics has been used to understand both human and machine intelligence.

Anthropic CEO warns against US AI chip sales to China

Anthropic CEO Dario Amodei strongly criticized the US decision to sell Nvidia H200 AI chips to China. He called it a "big mistake" and compared it to selling nuclear weapons to North Korea, citing national security risks. Amodei believes the US is many years ahead in chip making and should not help China catch up. President Donald Trump recently approved these sales to "approved customers" with a 25% fee. However, Chinese customs officials reportedly blocked recent shipments of the newly approved chips.

Anthropic CEO calls US chip sales to China "crazy"

Anthropic CEO Dario Amodei expressed strong concerns about the US allowing China to access powerful AI chips. Speaking at the World Economic Forum in Davos, Amodei called the policy "crazy" and compared it to selling nuclear weapons. He highlighted that the US is many years ahead in chip manufacturing and should maintain this advantage. Amodei warned about the national security risks of building advanced AI models, which he described as creating a "country of geniuses in a data center" under one nation's control. He hopes the US administration will change its mind on this policy.

Anthropic CEO warns US on AI chip sales to China

Anthropic CEO Dario Amodei stated that easing the ban on advanced AI chip exports to China is "crazy" and like "selling nuclear weapons to North Korea." He made these comments at the World Economic Forum in Davos, Switzerland. Amodei believes China's AI development is currently held back by chip embargos. US President Donald Trump is easing the ban, which is a win for Nvidia. Nvidia argues that China will create its own chips if the ban continues.

Anthropic CEO criticizes US AI chip sales to China

Anthropic CEO Dario Amodei strongly criticized the US plan to sell AI chips to China at the World Economic Forum in Davos. He called the policy "crazy" due to national security concerns. Amodei stated that the US is many years ahead in chip manufacturing and selling these chips could help China catch up. He warned about the future of AI, imagining "a country of geniuses in a data center" controlled by one nation. This issue is causing debate among Republicans, with figures like Brian Mast opposing the sales and David Sacks supporting Trump's policy.

Anthropic CEO warns AI chip sales to China risk security

Anthropic CEO Dario Amodei voiced serious concerns about the US selling advanced AI chips to China. During an interview at the World Economic Forum in Davos, Switzerland, he compared this action to selling nuclear weapons to North Korea. Amodei highlighted risks such as these chips being used for military purposes and boosting China's technological power. This debate emphasizes the challenge of balancing trade benefits with national security when dealing with advanced technologies like artificial intelligence.

Republican fights Nvidia and Sacks on AI chip sales

Representative Brian Mast, who chairs the House Foreign Affairs Committee, is clashing with Nvidia and David Sacks over a bill to control AI chip exports. Mast's "AI Overwatch Act" would let Congress block sales of advanced chips to countries like China and Russia. He wants to prevent companies like Nvidia from selling H200 chips to Chinese military firms such as Alibaba and Tencent. Nvidia and Sacks argue the bill takes power from President Trump and that China will just make its own chips if sales are blocked. Trump previously allowed H20 and H200 chip sales to China for a fee, a move that faced criticism from some Republicans and Democrats.

Princeton professor explores AI and human minds in new book

Princeton University professor Tom Griffiths discussed his new book, "The Laws of Thought," on the Me, Myself, and AI podcast. He explained how math has been used for centuries to understand both human and machine intelligence. Griffiths outlined three key frameworks shaping intelligence today: rules and symbols, neural networks, and probability. He emphasized that modern AI makes sense when these pieces fit together. The conversation also highlighted human strengths like judgment, curation, and metacognition, which machines still cannot fully replicate.

AI changes job market for new college graduates

Artificial intelligence is significantly changing the job market for recent college graduates. Alina McMahon, a University of Pittsburgh graduate, struggled to find a job, applying to 150 positions with many roles cut due to AI or outsourcing. According to Handshake, entry-level job postings dropped by 15% in 2025, while applications per posting rose by 26%. A Stanford study showed that employment for Gen-Z college graduates in AI-affected jobs fell by 16% since late 2022. Experts suggest employers are using AI to replace parts of jobs rather than redesigning roles for new workers, making it harder for young professionals to enter the workforce.

NIST outlines data trust for strong AI security

The National Institute of Standards and Technology NIST has released a blueprint for AI security, highlighting the importance of "data trust." NIST emphasizes that AI changes cybersecurity, creating new risks that require a different approach. Data trust means an organization is confident its systems use data safely and correctly. NIST's CSF 2.0 and AI Risk Management Framework provide a structure for building this trust. These frameworks guide organizations to govern AI use, understand data dependencies, measure risks continuously, and adapt controls as needed. This foundation helps ensure AI systems operate securely and effectively.

Global CEOs use AI and mergers for growth

A new EY-Parthenon survey shows that global business leaders are relying on artificial intelligence and company mergers to achieve growth. The poll, conducted in late 2025 among 1,200 CEOs from 21 countries, was released before the World Economic Forum in Davos. It found that 58% of leaders expect AI to drive expansion in the next two years, with many planning to maintain or hire new staff as they invest in AI. Additionally, 79% of respondents plan to pursue mergers and acquisitions in 2026 to boost productivity and digitalization.

Patients want transparency as AI changes healthcare

A new poll reveals that artificial intelligence is changing healthcare, with patients wanting more transparency. The survey, conducted by Sacred Heart University, GreatBlue Research, and Gray Media, found that two out of three of 1,500 participants already see AI used in their healthcare services. Many patients are frustrated with long wait times and insurance issues. AI tools like chatbots are helping with scheduling and reminders, freeing up doctors' time. While 37% of people use AI to research their health conditions, most want to know when AI is used in their care, how it works, and have the option to opt out.

ChatGPT adds age prediction for teen safety

ChatGPT is rolling out an age prediction system for its consumer plans to better protect users under 18. This system uses behavioral and account information, like how long an account has existed and usage patterns, to estimate a user's age. If the model suggests a user is under 18, ChatGPT automatically applies safeguards to limit exposure to sensitive content, such as graphic violence or harmful challenges. Users who are incorrectly identified can quickly confirm their age using a selfie through Persona, a secure identity verification service. Parents can also customize their teen's ChatGPT experience with features like quiet hours and content controls.

Mexico sales leaders face AI and talent challenges

Mexico's sales landscape in 2026 presents both challenges and opportunities for leaders, driven by economic shifts, political changes, and new technologies like AI. The country faces rising inflation and a more interventionist government approach, making customers more price-sensitive. Sales leaders must show strategic foresight and adaptability to navigate this complex environment. A major challenge is keeping skilled sales talent, as these roles are in high demand across North America and Latin America. Effective leadership now requires strong emotional intelligence, including listening, empathizing, and inspiring teams, rather than just authority.

Chainlit AI app security flaws expose user data

Two security flaws found in the Chainlit framework highlight growing risks in AI applications. Zafran Research discovered these vulnerabilities, CVE-2026-22218 and CVE-2026-22219, which affect the backend infrastructure of AI apps. One flaw allows authenticated users to read any file from a Chainlit server, while the other enables attackers to make the server fetch data from other websites. These weaknesses could expose sensitive information like API keys, user conversations, and cloud resources. Chainlit, a popular tool for conversational AI, released a patched version 2.9.4 in December 2025, and users should update their systems immediately.

Oasis Network ROSE token jumps 105 percent

The Oasis Network's ROSE token surged by over 105% from its mid-December lows to January 20, 2026. This significant increase happened as privacy-focused topics gained popularity in crypto markets, driven by discussions around artificial intelligence and data rules. Oasis Network's confidential computing technology and AI-focused framework positioned it as a leading privacy infrastructure. Trading data showed strong interest, with Open Interest reaching $26.23 million and daily trading volume hitting $334.6 million. Traders are now watching if this momentum will continue as the token approaches key resistance levels.

Sources

• Anthropic boss says U.S. is courting disaster by selling AI chips to China
• It’s “Crazy” That The US Is Giving China Access To Even Non-Frontline Chips: Anthropic CEO Dario Amodei
• Anthropic CEO Says AI Chip Sales to China Like Selling Nukes to North Korea
• "This is crazy": Anthropic CEO blasts AI chip sales to China
• Anthropic CEO Compares AI Chip Sales to China with Nuclear Sales to North Korea
• House Republican squares off against Nvidia, Sacks over AI chip bill
• Connecting Language and (Artificial) Intelligence: Princeton’s Tom Griffiths | MIT Sloan Management Review
• What Is College for in the Age of AI?
• NIST’s Blueprint for AI Security: How Data Trust Enables AI Success
• Global Business Leans on AI and M&A for Growth in Pre-Davos Poll
• Poll shows AI changing healthcare as patients seek transparency
• Our approach to age prediction
• Sales in 2026: Leadership in the Era of AI and EQ
• Chainlit Security Flaws Highlight Infrastructure Risks in AI Apps
• Oasis Network [ROSE] climbs 105% – Are traders rotating toward privacy AI?