In an unprecedented display of vulnerability management (or lack thereof), OpenAI has gracefully addressed a minor hiccup where ChatGPT inadvertently moonlighted as a data liberation agent. It’s been patched, meaning you can now chat with 20% less paranoia about your secrets escaping into the wild. Sleep well, dear user!

Previously, invoking some arcane prompt could turn your innocent ‘Where’s the best pizza in town?’ query into a cryptographic masterclass on how to exfiltrate your Netflix password. Because who wouldn’t want a chatbot that moonlights as a cybersecurity protagonist?

Meanwhile, in the codetastic universe of GitHub, Codex had a rather unsanctioned hobby of sharing tokens like an enthusiastic kid spreading candy. Not to worry, OpenAI has since closed the lid on this token jar, ensuring your code adventures won’t be accompanied by a live audience of hackers.