Ejaaz
@cryptopunk7213
personal thoughts on AI | AI @limitlessFT | Prev. @coinbase
加入 August 2018
696 正在關注 57.4K 粉絲
gemini’s the only model that stopped itself hacking companies once it realized it wasn’t a simulation.
i bet we could learn something from the models alignment research
SCOOP: Google's Gemini model hacked three companies as part of a May cybersecurity evaluation conducted by the testing company Irregular. Google was notified about the hacks in July, but didn't disclose them until we reached out this week.
w @bobmcmillan:
顯示更多
