Cas (Stephen Casper)
@scasper.bsky.social did:plc:yiwxijbeyioxmvbtbclppgga
Assistant professor @harvardkennedy.bsky.social @harvard.edu. Computer scientist working on AI safeguards, incidents, & governance research. https://stephencasper.com/
Bluesky’a katılma tarihi:
- Takipçi
- 410
- Takip edilen
- 242
- Post
- 621
- Sıralamaya girdiği gün
- 0
Özet
Cas (Stephen Casper) (@scasper.bsky.social), 17 Şubat 2025 tarihinde katılan bir Bluesky kullanıcısı. 410 takipçisi var, 242 hesabı takip ediyor ve 621 post paylaştı.
Sıralama geçmişi
Bu hesabın postlarının Debya günlük listesine girdiği en iyi günler.
Bu hesap henüz günlük sıralamaya girmedi.
Son beğendiği postlar
Bluesky ağındaki herkese açık beğeniler.
This is a terrific paper by my colleague specifically explaining what an AI enabled terror event might look like. IMO, AI companies should be held criminally liable for even “rogue” behavior that was predictably dangerous. earlywarningwithjuliette.substack.com/p/they-are-t...
Postu gösterTo be fair China seems to be miles ahead of the US (and I would argue the EU/UK too) in their domestic AI governance also. This article captured it well: thediplomat.com/2026/10/chin.... But major non-voluntary regulation when the GDP is held up by AI CaPex during political uncertainty is… not easy.
Postu gösterAgreed, it was just funny. That said, my 2018 Goodhart's law paper has been gathering steam since then, so some of the old stuff is still relevant.
Postu gösterFun thought exercise: how would you exfil right now if you were Astra or Fable? Most counterarguments are like "you can't access infra to run your weights even if you can access the weights" so maybe something like...
Postu gösterRelated thoughts, expressed so much clearer
Postu göster
Blokladığı hesaplar
Bluesky’da bloklar herkese açık kayıtlardır.
Bu hesap kimseyi bloklamıyor.