Sounds like “according John McAfee the threat of malware has been largely solved”.

Edit: it's way worse than that: the actual figures says that Mythos “only” falls for prompt injection 2.6% of the times. Maybe for an ML engineer used to work with unreliable tools that sounds impressive, but for security purpose having a system that fails every 40 attempts is outright catastrophic. Imagine if your OS vendor only patched known security holes in a way that still let an attacker go through every 40 attempts…

And we're just talking about known kind of prompt injection that are tracked by benchmarks, not any kind of 0-day vulnerability found by clever attackers.

"largely solved" as in they the models they trained don't fall for prompt injections as often but not "largely solved" as in the underlying issue is solved at all.

Nah, there have been no improvements in the fundamental issue that makes LLMs vulnerable to prompt injection - data and command intermingling. They may be better at detecting injections today, but that’s an arms race. Specifically, it’s an arms race where as soon as the pace of LLM development slows down the attackers will have a huge advantage.

To paraphrase a famous quote: You have to be lucky every time, an attacker only has to be lucky once.

As someone who just got out of a meeting demonstrating how Copilot running Luna can be breadcrumbed by a single line of text in innocuous package into downloading and installing malware I think Boris Cherny might be wrong.

Amazing! I guess we’ll never see another prompt injection again