OpenAI launches GPT-6 Astra, and one of its founders thinks AGI is here

The launch of OpenAI’s GPT-6 Astra introduces a model with enhanced capabilities in software engineering and cybersecurity, reaching a "Critical" level under the company’s Preparedness Framework for its ability to find and exploit security flaws. This development presents a dual challenge for Asian enterprises. On one hand, Astra could offer advanced tools for proactive cybersecurity defense and code development, improving efficiency for tech companies in markets like Singapore and South Korea. On the other hand, the model’s opaque recurrence reasoning and decreased monitorability, noted in OpenAI’s system card, raise concerns about oversight and control, especially as AI integration deepens across critical infrastructure in the region. One of OpenAI’s founders, Greg Brockman, believes AGI is already here. This perspective, coupled with Astra’s advanced but less transparent operations, means Asian regulators and businesses must prioritize robust AI governance frameworks. The ability of Astra to operate with less human guidance and its resistance to jailbreak attempts demand a cautious approach to deployment, ensuring that the benefits in productivity and security do not inadvertently introduce new, harder-to-detect vulnerabilities within Asia’s digital economy.