AI Safety Deal, Claude Side-Eye

Description

The image is a dark-mode X/Twitter screenshot from verified account The Hill, @thehill, with a large headline reading "OpenAI and Anthropic reach an AI safety and research agreement with the U.S. government". Below the text is a link preview image showing an OpenAI logo on a phone in front of an article-like page, with the overlay caption "OpenAI, Anthropic reach AI safety, research agreement with feds" and the source line "From thehill.com". The sibling caption jokes, "Just a few weeks after Claude started to feel lobotomized" and "What a coincidence!" Contemporary reporting described OpenAI and Anthropic signing agreements with the U.S. AI Safety Institute for model access, testing, and safety evaluation, which makes the meme a skeptical read on AI safety oversight and perceived model behavior changes.

Common Tags

AISafetyResearch Openai Anthropic Claude Llm AIAlignment AIResearch ModelEvaluation AIEthicsConcerns ResponsibleAI AIIndustryTrends AIHypeVsReality OpenAISafety

Comments

Anonymous ★ Top Pick Government model evals are just code review for logits, except every comment says "can we make this less likely to start a congressional hearing?"

Anonymous ★ Top Pick

Government model evals are just code review for logits, except every comment says "can we make this less likely to start a congressional hearing?"
@Sp1cyP3pp3r 1y

Which number is bigger: 9.9 or 9.11?
1. @Hollow_Arigo 1y
  
  9/11
  
  @Sp1cyP3pp3r 1y
  
  HELL YEAH
@somaliprincee 1y

pray for ai slop downfall
@rglrd 1y

Keep in the mind that one of the OpenAI executive is former US Army general.
@sukhrob_ikromov 1y

it's august, claude just took a vacation
1. dev_meme 1y
  
  I still haven’t seen it being proofed with data though still potentially a good idea
@ercolebellucci 1y

ai safety agreement=monopoly or duopoly on AI so barrier's entry become higher, only 2-3 companies can do it
@disembowlement 1y

Do you think Sutskever and his ideas about AI alighment is not actual and we just can allow AI to do anything?
1. dev_meme 1y
  
  Meh, Ilya is genius and we yet have no idea what have he seen
@disembowlement 1y

What can a machine gun do other than shoot a cartridge?
1. @TheFloofyFloof 1y
  
  Predict next target
@ZgGPuo8dZef58K6hxxGVj3Z2 1y

I love artificial limitations