September 17, 2025 News
AI Security Firm Irregular Secures $80M to Test and Secure Frontier AI Models Against Emergent Risks
AI security company Irregular raised $80 million led by Sequoia Capital to develop systems that identify emergent risks in frontier AI models before they are released. The company uses complex network simulations where AI agents act as both attackers and defenders to test model vulnerabilities and security weaknesses.
Skynet Chance (-0.08%): The development of robust AI security testing and vulnerability detection systems reduces the probability of uncontrolled AI deployment by creating better safeguards and early warning systems for dangerous capabilities.
Skynet Date (+0 days): Investment in AI security infrastructure may slightly slow deployment timelines as more rigorous testing becomes standard practice, though this represents a minor deceleration in the overall pace.
AGI Progress (+0.01%): The focus on securing increasingly sophisticated AI models indicates continued advancement in frontier model capabilities, and the security testing itself may contribute to understanding AI behavior and limitations.
AGI Date (+0 days): Enhanced security requirements and testing protocols may add minor delays to model development and deployment cycles, slightly decelerating the pace toward AGI achievement.
China Bans Domestic Tech Companies from Purchasing Nvidia AI Chips
China's Cyberspace Administration has banned domestic tech companies from buying Nvidia AI chips and ordered companies like ByteDance and Alibaba to stop testing Nvidia's RTX Pro 6000D servers. This follows previous US licensing requirements and represents a significant blow to China's tech ecosystem, as Nvidia dominates the global AI chip market with the most advanced processors available.
Skynet Chance (-0.08%): Restricting access to advanced AI chips could slow the development of the most capable AI systems in China, potentially reducing the overall global risk of uncontrolled AI development. However, this may also push China toward developing independent AI capabilities without international oversight.
Skynet Date (+1 days): The chip ban will likely delay China's AI development timeline by forcing reliance on less advanced local alternatives, potentially slowing the pace toward scenarios involving advanced AI systems. This deceleration effect is partially offset by the motivation for accelerated domestic chip development.
AGI Progress (-0.05%): Limiting access to the world's most advanced AI chips represents a significant setback for AGI development in China, as these chips are crucial for training large-scale AI models. This fragmentation of the global AI development ecosystem may slow overall progress toward AGI.
AGI Date (+1 days): The ban forces Chinese companies to use less capable hardware alternatives, which will substantially slow their AI research and development timelines. This represents a meaningful deceleration in the global race toward AGI achievement.