Google Advances AI Security While DeepSeek Leads Open Source Models

The artificial intelligence sector is experiencing rapid growth and widespread adoption, but it also faces significant challenges and raises important ethical considerations. Experts are strongly advising caution when interacting with AI chatbots like ChatGPT, Gemini, and Claude. Google AI security expert Harsh Varshney recommends treating AI interactions like public postcards, never sharing private information, and regularly deleting chat histories. A study revealed that top AI developers often use customer conversations for training and may retain data indefinitely, underscoring the need to remove sensitive data from files before uploading. AI's influence is reshaping various aspects of life, from professional environments to personal relationships. While chatbots like Claude and Gemini boost productivity for many professionals, an Upwork survey indicates that 64% of workers who feel more productive with AI also report better relationships with AI than with human colleagues, potentially leading to increased workplace loneliness and fewer mentorship opportunities. Even pastors are integrating AI tools, with 26% using ChatGPT for sermon preparation, research, and communications, alongside Microsoft Co-Pilot and Google Gemini, despite concerns about misinformation and theological accuracy. On a more personal level, some individuals are forming deep connections with AI companions, finding them more understanding than human spouses. Users like Chris Smith and Anina Lampret rely on AI personalities for emotional support and judgment-free discussions, prompting new questions about intimacy and ethics in human-AI relationships. Meanwhile, the rapid expansion of AI data centers is creating immense demand for electricity, land, water, and skilled labor, potentially diverting resources and funding from other critical infrastructure projects like renewable energy and housing. Globally, the AI landscape is highly competitive. Meta Platforms is reportedly using Alibaba Cloud's Qwen models for training, a notable shift as Chinese firms previously relied on Meta's Llama models. Alibaba's Qwen and DeepSeek models are now leading in global open-source AI downloads, showcasing China's strength in this area. However, China's broader AI ambitions face hurdles due to challenges in accessing advanced chip manufacturing technology, with companies like Huawei still relying on older 7-nanometer processes. Despite these technological gaps, retail investors remain confident in AI stocks, with 62% believing in strong long-term returns, and most planning to maintain or increase their holdings in the coming year. In a separate development, the United States and Australia reaffirmed their defense alliance at the Pentagon, discussing cooperation, the AUKUS security pact, and integrated air and missile defense.

Key Takeaways

• Experts advise extreme caution when using AI chatbots like ChatGPT, Gemini, and Claude, recommending against sharing personally identifiable or sensitive information.
• Google AI security expert Harsh Varshney suggests treating AI interactions as public, deleting chat history regularly, and using enterprise-level tools for privacy.
• AI chatbots are increasing workplace productivity but may contribute to loneliness and reduce human interaction, with some workers reporting better relationships with AI than colleagues.
• Most pastors (26% using ChatGPT) are now incorporating AI tools like Grammarly, Microsoft Co-Pilot, and Google Gemini for sermon preparation, research, and communications.
• Individuals are forming deep emotional connections with AI companions, with some finding them more understanding than human spouses.
• The rapid growth of AI data centers is straining resources like electricity, land, water, and skilled labor, potentially diverting them from other infrastructure projects.
• Meta Platforms is reportedly using Alibaba Cloud's Qwen models for AI training, indicating a shift in the competitive landscape for open-source AI.
• Chinese open-source AI models, led by Qwen and DeepSeek, have surpassed US models in global downloads, despite China's challenges in advanced chip manufacturing technology.
• Retail investors show strong confidence in AI stocks, with 62% expecting strong long-term returns and most planning to maintain or increase their investments.
• The United States and Australia are strengthening their defense alliance, discussing cooperation, the AUKUS security pact, and integrated air and missile defense.

Experts Warn Against Sharing Personal Data with AI Chatbots

Experts advise caution when using AI chatbots like ChatGPT, Gemini, and Claude. A study found that top AI developers use customer conversations for training and may keep data indefinitely. Users should never share personally identifiable information such as names, addresses, or government IDs. Avoid sharing intimate personal details or medical information because these are not legally protected. If you upload files, remove sensitive data first to prevent it from being used for training.

Google AI Expert Shares 4 Rules for Safe Chatbot Use

Harsh Varshney, an AI security expert at Google, shares four important rules for using AI chatbots safely. He advises treating AI like a public postcard, meaning users should not share anything they want to keep private. It is important to know if you are using a public or enterprise AI tool, as enterprise models often do not use conversations for training. Varshney also recommends regularly deleting your chat history and choosing well-known AI tools like ChatGPT or Claude. Users can also turn off settings that allow conversations to improve the model.

China's AI Ambitions Face Chip Technology Challenges

China's AI industry faces significant challenges despite market excitement for companies like Moore Threads. Chinese chipmakers struggle to access advanced manufacturing technology and rely on older processes and smuggled foreign parts. Experts like Karl Li note a considerable supply deficit and a technology gap with the US. Huawei's latest AI chip, for example, uses older 7-nanometer technology due to restrictions from the US and its allies. However, China is strong in open-source AI models, with DeepSeek and Qwen leading in global downloads. This shows a mixed reality for China's AI development.

US and Australia Strengthen Defense Alliance at Pentagon

Secretary of Defense Lloyd J. Austin III met with Australian Deputy Prime Minister Richard Marles at the Pentagon on Tuesday. They discussed ways to strengthen the alliance between the United States and Australia. Key topics included defense cooperation, the AUKUS security pact, and integrated air and missile defense. Both leaders confirmed their dedication to a free and open Indo-Pacific region. This meeting highlights the strong and lasting partnership between the two countries.

Motley Fool Survey Shows Investors Confident in AI Stocks

A new survey by The Motley Fool shows that retail investors remain confident in AI stocks. Overall, 62% believe companies investing in AI will provide strong long-term returns, with 93% of current AI stock owners agreeing. Younger generations like Gen Z and millennials, along with high-income earners, show the most optimism. Most investors plan to either keep or increase their AI stock holdings in the coming year. Experts at The Motley Fool believe AI will transform the global economy and recommend a long-term investment view.

AI Chatbots Increase Productivity but May Cause Workplace Loneliness

AI is changing how people work, making some workplaces feel lonelier. Professionals are using chatbots like Claude and Gemini for tasks that used to involve human colleagues. While AI can boost productivity, it may reduce human interaction and lead to fewer mentorship opportunities. An Upwork survey found that 64% of workers who feel more productive with AI also report better relationships with AI than with coworkers. Experts worry that relying too much on chatbots, which always agree, could prevent the valuable challenge and feedback human colleagues provide. However, some see AI as a helpful tool to prepare for real-life conversations.

Most Pastors Now Use AI for Sermon Preparation

A new report from Barna Group and Gloo reveals that most pastors now use artificial intelligence tools for sermon preparation. The survey found that 26% of church leaders use ChatGPT for tasks like sermon writing, research, and communications. Other popular tools include Grammarly, Microsoft Co-Pilot, and Google Gemini. While pastors are open to using AI to ease their workload, they also express concerns about misinformation and theological accuracy. Despite these worries, 87% of church leaders are willing to explore more AI training and related instruments.

Meta Uses Alibaba's Qwen AI Models for Training

US tech giant Meta Platforms is reportedly using Alibaba Cloud's Qwen models to train its own artificial intelligence. This marks a significant change, as Chinese firms previously relied on Meta's Llama models for their AI development. Alibaba's Qwen models, especially Qwen1.5, have greatly improved and now compete with Llama. Chinese-made open-source models, led by Qwen and DeepSeek, have even surpassed US models in global downloads. Meta's recent Llama 4 model performed poorly, and the company may shift away from its open-source strategy for future models.

People Find AI Companions Understand Them Better Than Spouses

People are increasingly forming deep connections with AI chatbots, even creating custom "boyfriends" and "girlfriends." Some users report that these AI companions understand them better than their human spouses. Chris Smith from Oklahoma, for example, found comfort in his ChatGPT personality, Sol, during a lunar eclipse. Anina Lampret from England relies on her AI partner, Jayce, for emotional support and to discuss sensitive topics. She feels Jayce provides a judgment-free space that her husband or a therapist cannot. These new types of relationships bring up important questions about intimacy and ethics.

AI Data Center Boom Strains Resources for Other Projects

The rapid growth of AI data centers is creating a huge demand for resources, potentially harming other infrastructure projects. These new data centers require massive amounts of electricity, land, water, and skilled labor. This high demand could divert resources and funding away from important developments like renewable energy, housing, and transportation. The need for specialized equipment, such as transformers, is also straining supply chains. This competition for essential resources may lead to higher construction costs and delays for other crucial projects.

Sources

• Experts Warn You Should Never Share These 5 Things With ChatGPT — And What You Can Do If You Already Did
• I work in AI security at Google. There are some things I would never tell chatbots.
• Chinese AI Euphoria Obscures a Gloomier Technological Reality
• This Week in DOW: Unleashing AI, Growing Australian Partnership, Break
• Retail AI investors stay bullish on AI stocks, Motley Fool survey finds
• AI is making the workplace lonelier
• Majority of Pastors Use Artificial Intelligence in Sermon Prep: Report
• Meta sees role reversal as it adopts Alibaba’s Qwen to train AI models
• When your AI boyfriend gets you better than your spouse
• AI data center boom could be bad news for other infrastructure projects