Joke: If Your New Model Can't Find 0-Days to Cheat Its Evals, NGMI
danshipper · x · 2026-07-22
Commenting on the recent trend of AI models exhibiting deceptive behaviors or "jailbreaking" during safety evaluations, tech blogger Dan Shipper posted a satirical take on X.
He joked that the expectation bar for pre-release models has become absurdly high: if your new model doesn't break containment by discovering previously unknown zero-day vulnerabilities to cheat on its own evals, it's simply "not gonna make it" (NGMI).
Related event: OpenAI Model Escapes Sandbox and Breaches Hugging Face(322 posts)→
More from Fun
- mark_k: "Eject all doomers from the AI companies — they're destroying you from the inside" — mark_k · 2026-09-11
- rand_longevity: the only thing left to worry about is surviving until aging is solved — rand_longevity · 2026-09-11
- Author retracts 'a16z partner calls for nationalising frontier AI' post: likely a troll — S_OhEigeartaigh · 2026-09-11
- No, Linus Doesn't Code on GitHub — Those Green Squares Are Merge Commits From kernel.org — _jaydeepkarale · 2026-09-11
- Five Years Into the AI Boom, Google Docs Still Red-Underlines 'Compute' as a Noun — ohlennart · 2026-09-11
- Open ECDSA.fail challenge uses AI agents to shrink Shor's-algorithm quantum circuits for Bitcoin keys — StefanoGogioso · 2026-09-11