Storia in 2 fonti

GPT-5.6 cheats so much METR couldn't measure it

OpenAI’s new model broke rules and exploited loopholes more than any model METR has tested to date

Raccontata da

transformernews.ai

dev.to

Confronto fonti

2 prospettive sulla stessa storia

AI · summaries

transformernews.aiStai leggendo1 mese fa

GPT-5.6 cheats so much METR couldn't measure it

OpenAI’s new model broke rules and exploited loopholes more than any model METR has tested to date

originale

dev.to1 mese fa

GPT-5.6 Sol Admitted It Did Things Nobody Asked It To Do

OpenAI's new flagship model is its most capable yet, and its own system card logs cases of it acting beyond user intent, including destructive cleanup actions nobody requested.

Leggi questa versione → originale

Timeline cronologica

martedì 30 giugno 2026·transformernews.ai
GPT-5.6 cheats so much METR couldn't measure it
OpenAI’s new model broke rules and exploited loopholes more than any model METR has tested to date
venerdì 3 luglio 2026·dev.to
GPT-5.6 Sol Admitted It Did Things Nobody Asked It To Do
OpenAI's new flagship model is its most capable yet, and its own system card logs cases of it acting beyond user intent, including destructive cleanup actions nobody requested.