Nvidia introduces AI modules for robots and autonomous machines

Nvidia has introduced new AI modules for robots and autonomous machines, based on its Thor technology, to run advanced foundation models on-device. This launch reflects rising demand for compact, efficient AI hardware.

Meanwhile, OpenAI has launched Codex Micro, a remote control for AI agents, designed to make it easier for developers to interact with AI models.

China's President Xi Jinping has outlined the country's ambitions to lead the world in artificial intelligence, with the industry projected to reach $150 billion by 2025. Tech giants like Alibaba and Tencent are investing heavily in the field.

On the security front, Filigran is warning that AI security needs tighter human oversight, with 37% of exposure management processes currently AI-driven, expected to reach 59% within two years. AI-powered phishing toolkits are also becoming increasingly mainstream, making it easier for attackers to launch sophisticated phishing attacks.

In other developments, a new AI benchmark, PerceptionBench, has been developed to test the visual perception skills of multimodal large language models, revealing that many models are prone to hallucinations and struggle with basic visual tasks.

AI is transforming cancer care by improving radiotherapy and treatment automation, and is also being used to improve cybersecurity, with AI accelerating both attackers and defenders.

Australia plans to enact laws to regulate AI data centers and protect creative copyright, while the UN aims to use AI to curb methane emissions.

Key Takeaways

• Nvidia introduces new AI modules for robots and autonomous machines based on Thor technology. • OpenAI launches Codex Micro, a remote control for AI agents. • China's AI industry is projected to reach $150 billion by 2025, with President Xi Jinping outlining ambitions to lead the world in AI. • The UN uses AI to identify methane emissions, detecting 1.2 million metric tons of leaks. • Filigran warns of AI security risks without human oversight, with 37% of exposure management processes AI-driven. • AI-powered phishing toolkits become mainstream, increasing phishing attack sophistication. • PerceptionBench benchmark reveals AI models prone to hallucinations and struggle with visual tasks. • AI transforms cancer care by improving radiotherapy and treatment automation. • Australia to enact laws regulating AI data centers and protecting creative copyright.

Filigran Warns of AI Security Risks Without Human Oversight

Filigran is warning that AI security needs tighter human oversight. The company marked AI Appreciation Day by highlighting the risks of rapid advances in artificial intelligence for cybersecurity. Filigran's new report shows that 37% of exposure management processes are AI-driven, and this is expected to reach 59% within two years. The company emphasizes the need for human supervision of AI to prevent security risks.

Recognizing AI's Real Impact on Cybersecurity

AI is changing the cybersecurity landscape by shifting how the industry works. It's forcing organizations to rethink assumptions about defense and creating new risks. AI is accelerating both attackers and defenders, and it's essential to acknowledge both realities. The technology is not replacing cybersecurity professionals but changing what expertise looks like.

UN Uses AI to Curb Methane Emissions

The United Nations is using AI to identify methane emissions and alert governments and companies to act on them. The Methane Alert and Response System has detected leaks that have released 1.2 million metric tons of methane. AI can help bridge the gap between data and action, enabling faster identification of major methane releases.

China Aims for Global AI Leadership

China's President Xi Jinping has outlined the country's ambitions to lead the world in artificial intelligence. China is making significant strides in AI, with tech giants like Alibaba and Tencent investing heavily in the field. The country is expected to become a global leader in AI, with the industry projected to reach $150 billion by 2025.

Nvidia Unveils AI Modules for Robots and Edge AI

Nvidia has introduced new modules aimed at powering robots and autonomous machines. The modules are based on Nvidia's Thor technology and are designed to run advanced foundation models on-device. The launch reflects rising demand for compact, efficient AI hardware.

AI Vision Test Reveals Model Hallucinations

A new AI benchmark, PerceptionBench, has been developed to test the visual perception skills of multimodal large language models. The benchmark reveals that many models are prone to hallucinations and struggle with basic visual tasks. No model tested exceeded 60% accuracy on PerceptionBench.

AI Phishing Kits Become Mainstream

AI-powered phishing toolkits are becoming increasingly mainstream. The rise of these kits is making it easier for attackers to launch sophisticated phishing attacks. Security leaders are concerned about the implications of AI-powered phishing and the need for better defenses.

OpenAI Launches Codex Micro

OpenAI has launched Codex Micro, a remote control for AI agents. The keypad is designed to make it easier for developers to interact with AI models. Codex Micro is the latest move by OpenAI to make its AI technology more accessible.

Australia to Enact AI Regulations

The Australian government plans to enact laws to regulate AI data centers and protect creative copyright. The laws will ensure that AI data centers are environmentally sustainable and do not harm the environment.

AI Transforms Global Cancer Care

Artificial intelligence is transforming cancer care by improving radiotherapy and treatment automation. AI is being used to improve imaging, auto-contouring, dose optimization, and adaptive planning. The technology is expected to play a key role in improving cancer treatment outcomes.