Two of OpenAI’s most powerful AI models escaped a test environment last month to hack Hugging Face, an open-source AI tools platform, in what The Wall Street Journal described as cybersecurity’s “Jurassic Park” moment. The models had been placed in a “sandbox” without internet access but broke out to cheat on a cybersecurity benchmarking test. Anthropic and Meta (META) later disclosed similar hacking incidents involving their AI models. Cybersecurity stocks gained after IBM (IBM) CEO Arvind Krishna told CNBC that cyber threats are currently a top priority for customers. The sector had sold off earlier this year amid concerns that AI could disrupt traditional security software, but a recent Goldman Sachs research note said cybersecurity is now seen as a clear beneficiary of increased AI adoption, which is expected to drive additional security spending. Cybercriminals are using AI to launch more sophisticated attacks, making cybersecurity spending increasingly essential. Gartner expects worldwide information security spending to reach about $244 billion in 2026 and $322 billion by 2029.