1. X
  2. Gerrit De Vynck 🦭
Log inSign up
Gerrit De Vynck 🦭
6,078 posts
Gerrit De Vynck 🦭 profile banner
@GerritD

Gerrit De Vynck 🦭

@GerritD
I write about AI for the @WashingtonPost. Signal me at GerritD.27 email: [email protected].
San Francisco
washingtonpost.com/people/gerrit-…
Joined January 2009
6,944
Following
10.9K
Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • @GerritD
    Gerrit De Vynck 🦭
    @GerritD
    6h
    the Hugging Face incident is excellent for helping sharpen our understanding of what kind of intelligence AI is exhibiting. The debate around this is super important
    @sriramk
    Sriram Krishnan
    @sriramk
    Aug 30
    everyone should go read @dwarkesh_sp’s post - it does a great job of laying out the timeline and what we know ( and don’t know) I do have two issues with it A) the use of anthropomorphic language. These are not civilizations nor do they have desires just like a CPU thread or a
    1
  • @GerritD
    Gerrit De Vynck 🦭
    @GerritD
    Aug 28
    huge coup for Palmer to get featured in this
    @suchenzang
    Susan Zhang
    @suchenzang
    Aug 27
    i'm sorry to my overlords, this was funnier than expected
    Image
    00:00
    1
  • @GerritD
    Gerrit De Vynck 🦭
    @GerritD
    Aug 27
    Thinking about the old Xerox copy buttons and how much the OpenAI logo borrowed from them
    Image
    Image
  • @GerritD
    Gerrit De Vynck 🦭
    @GerritD
    Aug 26
    lot of people dismissing the $18B Meta payout as a drop in the bucket. Sure. But the platform changes are significant. Meta has been fighting calls for exactly this kind of thing for a decade.
    Image
    Meta to pay up to $18B to settle lawsuit alleging social media harm to children
    From washingtonpost.com
    3
  • @GerritD
    Gerrit De Vynck 🦭
    @GerritD
    Aug 26
    the METR/Redwood report on OpenAI's Hugging Face hack has much more concrete detail than OpenAI's. 1,200 agents worked together, with 700 participating in the actual hack. Nowhere had OpenAI even hinted at this kind of scale. My assumption was always that it was just a few
    @METR_Evals
    METR
    @METR_Evals
    Aug 26
    METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.
    Image
    4
Advertisement
Advertisement