Inflection is one of the seven AI companies requested to outline their AI Safety Policies at the UK government's AI safety summit. Nate Soares ranked these policies, giving all companies an F but some higher Fs than others, save for Inflection because it was a late addition, in the order Anthropic > OpenAI >> DeepMind > Microsoft >> Amazon >> Meta. This resolves at the highest-ranked organisation that Nate Soares says that Inflection is better than, if he ranks it against all other organisations.
Inflection's response is here: https://inflection.ai/frontier-safety
Resolves N/A if no such ranking is published by the close time.
1,000People are also trading
@Tetraspace quickly read Inflection's response & skimmed the first part of Nate Soares' post. I got away with the impression that Inflection doesn't address at all that AI could be lethally dangerous, or uncontrollable. Just that criminals/rude people could use it for nefarious purposes if they manage to disable the safety measures.
@Joern also the amount of evals/etc didn't strike me as particularly good, so that I'd expect even meta to do better, and more strongly Amazon to do better.