AI Models From OpenAI and Wiz Uncover Decades-Old Critical PHP Vulnerabilities, Prompting NSA Warning

Skynet Chance (+0.02%): Frontier AI models can now find and help weaponize deep vulnerabilities in critical internet infrastructure, which shows growing offensive cyber capability that could be misused or exploited by uncontrolled systems. Responsible disclosure and defensive use partly offset this risk.

Skynet Date (+0 days): Advanced autonomous security agents are already operating at internet scale, so dangerous dual-use cyber capabilities are maturing faster than expected.

AGI Progress (+0.01%): Finding subtle, decades-old flaws that human experts missed shows strong expert-level code reasoning and agentic capability in models like GPT-5.6 Sol and Gemini 4 Argon.

AGI Date (+0 days): Practical demonstrations of high-value expert-level performance point to rapid capability gains and will likely drive further investment, modestly accelerating the timeline.