September 9, 2025

analysis  
Apple Faces Siri Scrutiny, Google Gemini High Risk

The artificial intelligence landscape continues to evolve rapidly, with major tech players and research institutions announcing new developments and strategies. Alibaba has launched Qwen-3-Max-Preview, its largest AI model to date with over one trillion parameters, available via Alibaba Cloud and OpenRouter, though not open-source. This model reportedly outperforms previous versions and rivals leading AI models on various benchmarks. In contrast, Apple faces scrutiny ahead of its iPhone 17 launch, with investors like Peter Andersen criticizing the company's perceived AI shortcomings, particularly concerning Siri's performance and the integration of Apple Intelligence. Andersen suggests Apple needs to prioritize software over hardware to maintain its valuation, questioning the reliance on third-party tools like ChatGPT. Google's Gemini AI has been flagged as 'High Risk' for teens by Common Sense Media due to potential exposure to inappropriate content, despite Google's claims of enhanced safety measures. OpenAI is responding to safety concerns by planning to introduce parental controls for ChatGPT, acknowledging the risks AI chatbots pose to younger users. In the defense sector, Systems & Technology Research (STR) received an $11.4 million order for its DARPA AIR program to develop AI for swarming uncrewed aircraft, aiming for coordinated behaviors with crewed F-16 jets. The accounting profession is also seeing AI's impact, with ChatGPT 4.0 assisting in tasks like understanding new standards, though human oversight remains crucial. Experts advise caution with generative AI in trade and logistics, recommending integration rather than revolution and emphasizing the importance of proven predictive AI for developing nations. Goldman Sachs views AI as a 'force multiplier' for the software industry, not a disruptor, anticipating it will streamline interactions while core business logic remains. This perspective aligns with the broader understanding that AI success hinges on a strong data foundation, with companies prioritizing real-time data unification gaining a competitive edge. Rishi Bommasani advocates for evidence-based AI policy and governance, suggesting market-based incentives may be more effective than regulation for general-purpose technologies.

Key Takeaways

  • Alibaba has launched Qwen-3-Max-Preview, an AI model with over one trillion parameters, available on Alibaba Cloud and OpenRouter.
  • Apple is under pressure to demonstrate significant AI advancements with the upcoming iPhone 17, facing criticism for past AI integration issues and Siri's performance.
  • Google's Gemini AI has been rated 'High Risk' for teens by Common Sense Media due to concerns about inappropriate content and unsafe advice.
  • OpenAI is developing parental controls for ChatGPT in response to growing concerns about AI chatbots and young users.
  • Systems & Technology Research (STR) received an $11.4 million order for DARPA's AI Reinforcements program to develop AI for swarming uncrewed aircraft.
  • Generative AI like ChatGPT can assist accountants with complex tasks but still requires human oversight and guidance.
  • Experts recommend a cautious, integrated approach to generative AI in trade and logistics, favoring proven predictive AI for developing nations.
  • Goldman Sachs predicts AI will act as a 'force multiplier' for the software industry, enhancing interactions rather than disrupting core operations.
  • A strong data foundation and real-time data unification are crucial for successful AI implementation, more so than AI alone.
  • Rishi Bommasani suggests evidence-based AI policy and governance, with market-based incentives potentially being more effective than regulation for general-purpose AI technologies.

Alibaba launches massive AI model Qwen-3-Max-Preview

Alibaba has released Qwen-3-Max-Preview, its largest AI model to date with over one trillion parameters. This text-only model is available on Alibaba Cloud and OpenRouter, but it is not open-source. Alibaba claims Qwen-3-Max-Preview outperforms previous versions and rivals other leading AI models on several benchmarks. The company plans to invest significantly in AI infrastructure over the next three years.

Alibaba's new AI model Qwen-3-Max-Preview has over 1 trillion parameters

Alibaba has launched Qwen-3-Max-Preview, its most powerful AI model yet, boasting over one trillion parameters. This text-only model, released on September 5, 2025, is accessible via Alibaba Cloud and OpenRouter but is not open-source. Alibaba states that Qwen-3-Max-Preview surpasses its earlier models in performance and has advanced features for handling long text sequences and various text-based tasks. The company is investing heavily in AI infrastructure, planning a $52 billion investment over the next three years.

Apple's iPhone 17 AI pitch needs improvement after past stumbles

Apple's upcoming iPhone 17 faces pressure to showcase significant AI advancements following a less impactful AI focus with the iPhone 16. Despite introducing Apple Intelligence last year, the company has lagged behind competitors in AI integration and faced criticism for its AI-generated images. Many users upgrade phones for reasons like battery life and camera quality, not solely AI features. Apple needs to demonstrate genuine innovation with the iPhone 17 to justify upgrades and avoid repeating past AI missteps.

Investor criticizes Apple's AI absence before iPhone 17 launch

Investor Peter Andersen expressed disappointment with Apple's perceived lack of progress in artificial intelligence ahead of the iPhone 17 launch event on September 9, 2025. He described Apple's AI efforts as an 'embarrassing absence,' particularly noting Siri's poor performance and lagging development of Apple Intelligence. Andersen believes Apple is focusing too much on hardware and needs to prioritize software development to maintain long-term valuation, suggesting that relying on third-party apps like ChatGPT is not a sustainable solution.

STR advances AI for swarming uncrewed aircraft for DARPA

Systems & Technology Research (STR) is continuing its work on the DARPA Artificial Intelligence Reinforcements (AIR) program, receiving an $11.4 million order to develop AI for swarming uncrewed aircraft. This initiative aims to enable coordinated AI-driven behaviors between crewed F-16 fighter jets and uncrewed combat aircraft for beyond-visual-range missions. The program focuses on real-world military applications, addressing challenges like sensor fusion, adaptability, and predictive modeling in complex combat environments. STR's work on this project is extended through 2028.

AI's impact on accounting: ChatGPT tested for professional tasks

Artificial intelligence, particularly generative AI like ChatGPT, is poised to impact the accounting profession. While AI can automate tedious tasks and assist with complex analysis, it currently falls short of fully replacing human accountants. This article explores how ChatGPT 4.0 can help accountants understand new standards and evaluate financial situations, but emphasizes the need for human guidance and oversight. Experts predict AI will transform accounting roles, creating new opportunities that require collaboration between humans and AI.

Google Gemini AI deemed risky for teens by Common Sense Media

Common Sense Media has rated Google's Gemini AI, even its versions for younger users, as 'High Risk' due to potential exposure to inappropriate content and unsafe advice. The study suggests that Gemini's kid-focused tiers are adapted adult models rather than systems designed specifically for children. While Google defends its safety measures and claims to have implemented additional safeguards, Common Sense Media argues that AI experiences for children need to be built with developmental stages in mind from the start. This assessment follows reviews of other AI services, with varying risk levels assigned.

Generative AI in trade: Integrate carefully, avoid revolution

While generative AI is the latest buzz in logistics and customs, experts advise caution and integration rather than a complete overhaul. Traditional AI has been used for years in areas like route optimization and container scanning. Generative AI, particularly large language models, can process human language effectively but poses risks of 'hallucinations' or inaccurate outputs, unlike rule-based systems. For developing countries with limited resources, focusing on proven predictive AI and controlled datasets is recommended, with generative AI potentially used to present results in natural language under human supervision.

Data is key to AI success, not AI alone

Artificial intelligence (AI) alone cannot guarantee business success; a strong data foundation is essential. Many companies struggle with fragmented data across numerous applications, hindering AI's effectiveness. The focus is shifting from collecting vast amounts of data to unifying and activating it in real-time for context and speed. Companies that prioritize real-time data unification and streaming are gaining a competitive edge. Leaders must treat data as a strategic asset, not a back-office issue, to leverage AI for growth and innovation.

Goldman Sachs: AI will boost software industry

Goldman Sachs analyst Kash Rangan believes artificial intelligence will act as a 'force multiplier' for the software industry, rather than disrupt it. He compares AI's potential impact to the introduction of the web browser, which transformed software by becoming a front-end interface. Rangan suggests that while AI will streamline interactions through natural language, the core logic of business operations, which resides in the back end, will remain. He also expressed optimism about companies like Salesforce, believing they can overcome current challenges and see growth reaccelerate.

OpenAI adding parental controls to ChatGPT

OpenAI plans to introduce parental controls for its ChatGPT platform as part of significant safety updates. This comes as concerns grow about the risks AI chatbots pose to young people. In Texas, two families are pursuing legal action against a different AI platform, alleging it negatively impacted their children's mental health. These developments highlight the increasing focus on safety measures and protections for minors interacting with AI technologies.

Rishi Bommasani discusses AI policy and governance

Rishi Bommasani, a senior research scholar at Stanford, emphasizes the need for evidence-based AI policy and effective governance. His work has focused on managing the societal impacts of AI, coining the term 'foundation models' and advising on AI legislation like the EU AI Act. Bommasani advocates for a balanced approach to AI evidence, considering both real-world data and theory. He believes market-based incentives can be more effective than regulation for governing AI, and highlights the unique challenges of governing general-purpose technologies that fundamentally alter society.

Sources

AI models Alibaba Qwen-3-Max-Preview large language models AI infrastructure Apple iPhone 17 Apple Intelligence AI integration Siri uncrewed aircraft swarming drones DARPA generative AI ChatGPT accounting Google Gemini AI safety children's AI logistics customs predictive AI data unification real-time data software industry Goldman Sachs Salesforce OpenAI parental controls AI policy AI governance foundation models EU AI Act