Every AI escape on record was staged by the people running the test
The claim that a sealed superintelligence would talk its way out rests on unlogged roleplays its author calls anecdotal, and every self-exfiltration a real model performed ran on a server the test supplied.
One gradient step still fools image classifiers, a decade on
Goodfellow, Shlens, and Szegedy's 2015 paper blamed the panda-to-gibbon trick on models being too linear, a claim later work disputes, and proposed a defense that had to be rebuilt before it worked.
'Gay' filled 3% of toxic comments, so the classifier marked the word toxic
The number that ranks language models for toxicity is one classifier's probability, and the same labels score African-American English as more toxic.
An image generator never counts the fingers it draws
Hands are the worst case for a model that learns how images look but not how a hand is built, and better data is slowly closing the gap.
AI writing detectors flag non-native students' own essays as machine-written
Turnitin sold the tool to thousands of schools promising it would misread human writing as AI under 1 percent of the time, and students were accused of cheating on its flags.