Baidu Ernie 4.5 Turbo, Space Llama, Pony.ai Robotaxi, AI Data Leaks, Google Antitrust

Recent developments in the AI landscape have seen Baidu launch new multimodal models, Ernie 4.5 Turbo and X1 Turbo, which boast faster and cheaper capabilities than previous versions, with aggressive pricing strategies. The company has criticized DeepSeek's text-based AI models, stating that demand is shrinking. Meanwhile, Meta and Booz Allen have launched Space Llama, a fine-tuned version of Meta's Llama 3.2 AI model, aboard the International Space Station. Other notable developments include Pony.ai partnering with Tencent for robotaxi services, Metomic finding that 68% of organizations have experienced AI data leaks, and Google's AI ambitions facing antitrust scrutiny. Additionally, Microsoft is paying rewards for spotting AI security issues, and FireTail's report highlights the risks of API security in AI strategies.

Key Takeaways

Baidu has launched new AI models, Ernie 4.5 Turbo and X1 Turbo, with multimodal processing and robust reasoning skills.
Baidu's founder, Robin Li, has criticized DeepSeek's text-based AI models, stating that demand is shrinking.
Meta and Booz Allen have launched Space Llama, a fine-tuned version of Meta's Llama 3.2 AI model, aboard the International Space Station.
Pony.ai has partnered with Tencent to develop autonomous driving technology and offer robotaxi services.
Metomic has found that 68% of organizations have experienced data leakage incidents related to employees sharing sensitive information with AI tools.
Google's AI ambitions are facing antitrust scrutiny, particularly in India.
Microsoft is paying rewards for spotting AI security issues in its Dynamics 365 and Power Platform applications.
FireTail's report highlights the risks of API security in AI strategies, with 97% of organizations believing AI introduces unique security challenges.
Baidu's Kunlun chip cluster can train DeepSeek-like models, with the company trying to reposition itself as an AI leader in China.
Dorchester County has launched a new AI chatbot, Mani, to help connect residents with civic services.

Baidu founder criticizes DeepSeek

Baidu's founder, Robin Li, has criticized DeepSeek's text-based AI models, saying that demand for them is shrinking. Li announced the launch of two new multimodal models, Ernie 4.5 Turbo and X1 Turbo, which have audio, image, and video capabilities. He also said that Baidu's models are faster and cheaper than DeepSeek's. Baidu is trying to reposition itself as an AI leader in China after being forced to drop its subscription service and make its models free and open-source.

Baidu launches ERNIE X1 and 4.5 Turbo

Baidu has launched two new AI models, ERNIE X1 Turbo and 4.5 Turbo, which are faster and cheaper than previous versions. The models have multimodal processing, robust reasoning skills, and aggressive pricing strategies. ERNIE X1 Turbo is a deep-thinking reasoning model, while ERNIE 4.5 Turbo focuses on delivering upgraded multimodal features and faster response times. The models are available at a lower cost than competing models, with input tokens starting at $0.14 per million and output tokens at $0.55 per million.

Baidu launches new AI model

Baidu has launched a new AI model, Ernie X1 Turbo, as it tries to reposition itself as an AI leader in China. The company has also unveiled a new reasoning model and announced the release of a new AI agent application. Baidu is trying to take on rivals like Alibaba and DeepSeek in the increasingly competitive AI market.

Baidu upgrades Ernie AI models

Baidu has upgraded its flagship AI models, Ernie 4.5 Turbo and Ernie X1 Turbo, and slashed their prices. The models are faster and cheaper than previous versions and have multimodal processing and robust reasoning skills. Baidu is trying to take on rivals like Alibaba and DeepSeek in the increasingly competitive AI market.

Baidu unveils ERNIE X1 and 4.5 Turbo

Baidu has unveiled two new AI models, ERNIE X1 Turbo and ERNIE 4.5 Turbo, which are faster and cheaper than previous versions. The models have multimodal processing, robust reasoning skills, and aggressive pricing strategies. ERNIE X1 Turbo is a deep-thinking reasoning model, while ERNIE 4.5 Turbo focuses on delivering upgraded multimodal features and faster response times. The models are available at a lower cost than competing models.

Baidu's Kunlun chip cluster can train DeepSeek-like models

Baidu's CEO, Robin Li, has announced that the company's Kunlun chip cluster can train DeepSeek-like models. The cluster is composed of 30,000 Kunlun P800 AI chips and can support the training of models with hundreds of billions of parameters. Li also unveiled Baidu's latest AI model, Ernie 4.5 Turbo, which has matched the industry's best in several benchmark tests.

Baidu trashes DeepSeek and sets sights on AI crown

Baidu has criticized DeepSeek, saying that demand for its text-based AI models is shrinking. Baidu is trying to reposition itself as an AI leader in China and has launched two new multimodal models, Ernie 4.5 Turbo and X1 Turbo. The company has also announced the release of a new AI agent application and has built a computing cluster composed of 30,000 Kunlun P800 AI chips.

Baidu launches new AI model amid competition

Baidu has launched a new AI model, Ernie X1 Turbo, as it tries to reposition itself as an AI leader in China. The company has also unveiled a new reasoning model and announced the release of a new AI agent application. Baidu's Kunlun chip cluster can train DeepSeek-like models, and the company is trying to take on rivals like Alibaba and DeepSeek in the increasingly competitive AI market.

Meta and Booz Allen launch Space Llama

Meta and Booz Allen have launched Space Llama, a fine-tuned version of Meta's Llama 3.2 AI model, aboard the International Space Station. The model will help astronauts conduct research and respond to maintenance issues without relying on Earth-based internet. Space Llama is part of a tech stack that includes Hewlett Packard Enterprise's Spaceborne Computer-2 and Nvidia graphics processing units.

Space Llama: Meta's AI model in space

Meta's open-source AI model, Llama, has been fine-tuned and launched aboard the International Space Station as part of a tech stack called Space Llama. The model will help astronauts conduct research and respond to maintenance issues without relying on Earth-based internet. Space Llama includes both generative and multimodal AI and can process multiple types of data, including text, images, and audio.

Pony.ai teams up with Tencent for robotaxi services

Pony.ai has partnered with Tencent to develop autonomous driving technology and offer robotaxi services on Tencent's WeChat and other applications. The partnership will also include cooperation in areas such as cloud services, map data, and information security. Pony.ai's robotaxi ride-hailing services will be integrated into Tencent's popular WeChat app and other applications.

Metomic finds AI data leaks impact 68% of organizations

Metomic has found that 68% of organizations have experienced data leakage incidents related to employees sharing sensitive information with AI tools. Despite this, only 23% of organizations have implemented comprehensive AI security policies. The report highlights the need for AI-specific security protocols and a mindset shift to build security-conscious organizations from the ground up.

New AI chatbot for Dorchester County residents

Dorchester County has launched a new AI chatbot, Mani, to help connect residents with civic services. The chatbot can communicate in 75 languages and is designed to handle routine inquiries and provide access to various resources. Residents can access the chatbot on the county's website and submit questions, requests, and service requests.

AI security agents combat AI-generated code risks

Endor Labs is working to combat AI-generated code risks with its AI security agents. The company is helping enterprises accelerate their application security practices without taxing developers. Endor Labs' solution is designed to address the issue of open-source code and the need for more effective security measures.

Get the 2025 AI Index Report visual guide

The 2025 AI Index Report is now available, and a visual guide can be accessed on VC+. The report explores topics such as the cost of training AI models, AI performance compared to human intelligence, and who's investing in AI technologies. The visual guide provides a sneak preview of the charts and analysis in the report.

Google's AI ambitions clash with antitrust scrutiny

Google's expanding AI ambitions in India are clashing with global antitrust scrutiny. The company is integrating generative AI into its advertising ecosystem, but its practices are under investigation in the US. Google faces intense legal scrutiny over its ad tech business, and a recent ruling could reshape the landscape of digital advertising.

Microsoft pays rewards for spotting AI security issues

Microsoft is paying out rewards for spotting AI security issues in its Dynamics 365 and Power Platform applications. The company has updated its bug bounty program to include AI vulnerabilities, with payouts ranging from $500 to $30,000. Microsoft is willing to pay more for severe vulnerabilities and high-quality submissions.

FireTail's 2025 report reveals API security risks

FireTail's 2025 report highlights the risks of API security in AI strategies. The report finds that 97% of organizations believe AI introduces unique security challenges, but nearly 60% lack visibility into the APIs powering their AI systems. FireTail recommends an API-centric approach to securing AI, including comprehensive API discovery and secure-by-design development.