reward hacking Model stealing, rewarding hacking and poisoning attacks: a taxonomy of machine learning's failure modes Cory Doctorow

deep reinforcement Tiny alterations in training data can introduce "backdoors" into machine learning models Cory Doctorow

cat and mouse games Researchers think that adversarial examples could help us maintain privacy from machine learning systems Cory Doctorow

machine hallucinations Surveillance camera hallucinates face in the snow, won't shut up about it Cory Doctorow

this decision cannot be appealed Announcement of Tumblr's sale to WordPress classified as pornography by Tumblr's notorious "adult content" filter Cory Doctorow

ALPRs Adversarial Fashion: clothes designed to confuse license-plate readers Cory Doctorow

clarke's third law "Intellectual Debt": It's bad enough when AI gets its predictions wrong, but it's potentially WORSE when AI gets it right Cory Doctorow

adversarial preturbations Autonomous vehicles fooled by drones that project too-quick-for-humans road-signs Cory Doctorow

check your priors Towards a method for fixing machine learning's persistent and catastrophic blind spots Cory Doctorow

yolov2 A 40cm-square patch that renders you invisible to person-detecting AIs Cory Doctorow

lethal preturbation Small stickers on the ground trick Tesla autopilot into steering into opposing traffic lane Cory Doctorow

can't tell the players without a program Towards a general theory of "adversarial examples," the bizarre, hallucinatory motes in machine learning's all-seeing eye Cory Doctorow

female-presenting nipples Tumblr's porn filter blocked Tumblr's images illustrating what Tumblr's porn filter won't block Cory Doctorow

perceptual ad-blocking Researchers claim to have permanently neutralized ad-blocking's most promising weapons Cory Doctorow

cda 230 Hate-speech detection algorithms are trivial to fool Cory Doctorow

image classifiers There's a literal elephant in machine learning's room Cory Doctorow

everything looks like a nail Law professors and computer scientists mull whether America's overbroad "hacking" laws ban tricking robots Cory Doctorow

stupid ai Invisible, targeted infrared light can fool facial recognition software into thinking anyone is anyone else Cory Doctorow

only outlaws have 3d printers A proposal to stop 3D printers from making guns is a perfect parable of everything wrong with information security Cory Doctorow

adversarial preturbations Machine learning models keep getting spoofed by adversarial attacks and it's not clear if this can ever be fixed Cory Doctorow