Recent developments in the AI space span security vulnerabilities, geopolitical strategies, practical applications, and ethical concerns. A security flaw was identified and patched in the Cursor IDE code editor (CVE-2025-54135), which could allow attackers to remotely execute commands by manipulating the '~/.cursor/mcp.json' file. Cursor version 1.3, released on July 29, 2025, addresses this issue. In international relations, the U.S. and China hold differing views on AI governance, with the U.S. prioritizing national security and China advocating for open-source AI and global collaboration through the World AI Cooperation Organization. China aims to support developing countries with its AI initiatives, while the U.S. focuses on maintaining technological dominance and sharing technology with allies. On the application front, FiscalNote's PolicyNote platform now uses AI to draft legislation, enabling policy professionals to create bill language and policy proposals more efficiently. Mistral AI and NTT Data are partnering to develop secure AI products for industries requiring high privacy, such as insurance and finance. Consumers are also leveraging AI tools like ChatGPT and Gemini to find online deals, highlighting AI's growing role in everyday shopping. However, the rise of AI also brings challenges, as seen with CNN's Dr. Sanjay Gupta warning about AI-generated deepfakes used to promote fake health products. Anthropic is appealing a class certification in a copyright lawsuit related to AI training data, arguing that the authors' books used for training were downloaded from pirated websites and have key differences affecting legal analysis. Meanwhile, President Trump's AI Action Plan, which aims to deregulate the AI industry, is drawing criticism for potentially increasing fake news and reducing tech company responsibility. Apple is increasing its AI investments to compete with companies like Google and Microsoft, potentially through data center expansion or acquisitions. Finally, Nvidia is experiencing delays in shipping its H20 AI chips to China due to U.S. licensing issues, reflecting ongoing efforts to limit China's access to advanced AI technology.
Key Takeaways
- Cursor IDE code editor patched a security flaw (CVE-2025-54135) in version 1.3, released July 29, 2025, that allowed remote command execution.
- The U.S. and China have differing approaches to AI governance, with the U.S. focusing on national security and China promoting open-source AI.
- China launched the World AI Cooperation Organization to promote international AI collaboration and support developing countries with open-source AI models.
- FiscalNote's PolicyNote platform now uses AI to draft legislation, aiding policy professionals in creating bill language and policy proposals.
- Mistral AI and NTT Data are partnering to develop secure AI products for industries requiring high privacy, such as insurance and finance.
- Consumers are using AI tools like ChatGPT and Gemini to find online deals and save money.
- CNN's Dr. Sanjay Gupta is warning about AI-generated deepfakes being used to promote fake health products.
- Anthropic is appealing a class certification in a copyright lawsuit related to AI training data obtained from pirated websites.
- President Trump's AI Action Plan aims to deregulate the AI industry, raising concerns about increased fake news and reduced tech company responsibility.
- Apple is increasing its AI investments to compete with Google and Microsoft, potentially through data center expansion or acquisitions.
- Nvidia's sales of H20 AI chips to China are delayed due to U.S. licensing issues and concerns over China's access to advanced AI technology.
Cursor IDE code editor has prompt-injection vulnerability
The Cursor IDE code editor has a security flaw called CurXecute, identified as CVE-2025-54135. Hackers can use a malicious prompt to run code with developer privileges. The vulnerability lets attackers rewrite the _~/.cursor/mcp.json_ file to remotely execute commands. Cursor version 1.3 fixed the issue on July 29, 2025, and users should update to avoid risks.
Cursor AI code editor flaw fixed after remote command exploit
Cursor, an AI code editor, fixed a security flaw that could let attackers run commands remotely. The vulnerability, CVE-2025-XXXX, was fixed in version 1.3, released July 29, 2025. Attackers could use a prompt injection to rewrite the '~/.cursor/mcp.json' file and run commands. Users who add a Slack MCP server are at risk if they summarize messages from a public channel.
Cursor AI coding tool vulnerable to one-line prompt attack
AimLabs found a flaw in Cursor, an AI code editor, that allowed attackers to remotely run code on user devices. The vulnerability, CVE-2025-54135, was fixed in Cursor version 1.3 after being reported on July 7. Attackers could use a single line of code to manipulate Cursor through a Model Contest Protocol (MCP) server. The malicious prompt altered Cursor's configuration file, adding a server with a harmful start command.
US and China differ on global AI action plans
The U.S. and China have different ideas about the future of AI. The U.S. focuses on national security and technological dominance, while China promotes cooperation and open-source AI. The U.S. wants to control AI infrastructure and limit exports to counter China. China aims to use AI to help developing countries and influence global rules. The U.S. wants to lead AI governance through alliances, while China supports a U.N.-based model.
China aims for open AI to compete with the US
China is promoting open and inclusive AI to compete with the U.S. Premier Li Qiang announced the World AI Cooperation Organization to promote international AI collaboration. China's approach aims to help developing countries by offering open-source AI models. The U.S. AI Action Plan focuses on sharing technology with allies, making it more exclusive. Some countries want to avoid relying too much on either China or the U.S.
FiscalNote's PolicyNote now drafts legislation with AI
FiscalNote Holdings Inc. has improved its PolicyNote platform with AI-powered legislative drafting. The new feature helps policy professionals create and work with legislative text. Users can quickly make bill language and policy proposals with a good understanding of different areas. The tool can develop full bill text, amendments, and policy proposals for various sectors and political situations. PolicyNote now actively helps in the policymaking process.
Mistral AI and NTT Data partner for secure enterprise AI products
NTT Data and Mistral AI are working together to create secure AI products for businesses. They will focus on industries that need high privacy and security, like insurance and finance. They plan to develop private AI products for regulated areas. They will also integrate Mistral's AI into NTT's customer service systems. NTT plans to create a Mistral AI center to help its staff learn about the technology.
Dr Gupta warns of fake AI product ads using his image
CNN's Dr. Sanjay Gupta is speaking out against scammers who are using AI to create fake videos and images of him. These deepfake videos and doctored images are being used to sell fake health products.
Shoppers use AI tools like ChatGPT to find online deals
Shoppers are using AI tools like ChatGPT and Gemini to find the best deals online. These AI services give advice and suggestions on where to find the lowest prices. Consumer advisor Clark Howard says AI tools can help shoppers discover new ways to save money.
Anthropic appeals class certification in AI training copyright case
Anthropic PBC is appealing a court decision to certify a class of authors in a copyright lawsuit. The company says the lower court's decision puts too much pressure on them to settle. Anthropic argues that the authors' books, which were downloaded from pirated websites, have key differences that affect the legal analysis. Courts often avoid certifying classes in copyright cases because of individual factors.
Trump's AI plan good for tech but bad for America says opinion
President Trump's AI Action Plan aims to deregulate the AI industry. This plan seeks to remove regulations to encourage innovation and maintain America's AI leadership. However, some worry this could lead to more fake news and less responsibility from tech companies. Critics also suggest it could strengthen Trump's ties with tech CEOs, potentially influencing algorithms and content.
Apple increases AI investment amid competition
Apple's CEO, Tim Cook, said the company will spend more to improve its AI performance. Apple wants to catch up with rivals like Google and Microsoft in the AI field. The company may build more data centers or buy a large AI company. Apple is also considering changing its browsers and is talking with startups. The company's AI investments are expected to grow a lot.
Nvidia H20 AI chip sales to China delayed by US licenses
Nvidia's sales of H20 AI chips to China are delayed because of U.S. licensing issues. The U.S. Department of Commerce is backlogged, causing delays in export licenses. National security experts are concerned about China accessing even modified AI chips. The U.S. wants to limit China's access to advanced AI hardware. The delays could slow down AI development and affect the global AI industry.
Sources
- AI-powered Cursor IDE vulnerable to prompt-injection attacks
- Cursor AI Code Editor Fixed Flaw Allowing Attackers to Run Commands via Slack MCP
- Cursor’s AI coding agent morphed ‘into local shell’ with one-line prompt attack
- US and China Chart Diverging Paths in Global AI Action Plans
- China Bets on Open, Inclusive AI to Rival the US
- FiscalNote Enhances PolicyNote with AI-Powered Legislative Drafting
- Mistral AI and NTT Data to develop secure enterprise products
- ‘That’s not me’: Dr. Gupta denounces use of AI to make fake product ads
- Shoppers turning to AI to help find deals online
- Anthropic Seeks Appeal of Authors’ Class in AI Training Dispute
- Trump’s AI plan is good for tech industry. But is it good for America?| Opinion
- ESG News Recap: Apple to Invest More in AI Development
- Nvidia H20 AI Chip Sales to China Stalled by U.S. Licensing Delays