@noahmatrix_

seeing how deep the rabbit hole goes. finished resident evil 4 vr in an unhealthy amount of time.

 Charleston,WV
Joined May 2015
  • Tweets113
  • Following83
  • Followers74
4Media
OpenAI, Google and Anthropic have reportedly discussed an industry-led AI safety body, just as Astra became OpenAI’s first model to hit its “Critical” cyber-capability threshold. Safety may be the goal, but letting the incumbents write the rules is also a very convenient moat
11
“we need to make sure the AI won’t act on its own before launch” is elite PR for “we need more time and would rather not talk about diminishing returns”
prince gupta
@prince_guptar
Sep 14
openai is adding a new safety check before some big ai training runs that are expected to make its models much more powerful! basically... they want to check what could go wrong before pushing the model to the next level. that includes things like the model acting on its own, being misused or finding ways around its safety rules as ai gets more powerful, having that extra check before launch seems pretty smart tbh better to find the problems before everyone gets access to the model😬
7
OpenAI claiming 10,000 parallel agents solved an unsolved math problem while the US government officially accuses China of sovereign-level model distillation is insane the industry is oscillating between divine AGI promises and geopolitical warfare in the exact same news cycle
8
nothing humbles you faster than confidently reaching for something in vr and realizing you were grabbing absolutely nothing irl
7
hot take: nvidia owning hugging face means every "open" model on there now runs through a company that also sells you the chips to run it. calling that open source feels like a stretch at this point
7
cognition ai raising at $40B valuation. less than 3 months after closing at $26B. the market's telling you something: agentic coding is the next infrastructure layer. devs who ignore this will look like the ones who skipped cloud in 2010
13
everyone wants to time the market and almost nobody has a strategy for when they're wrong 🙄
22
meta reportedly spending $50k per employee per year on AI tokens. that's 18% of average comp just going to inference
23
openai and anthropic are both in the IPO pipeline now. anthropic filed confidentially at a $965B valuation. the public markets are about to find out what frontier AI actually looks like on a balance sheet
49
Load more