AI safety work criticized for being too theoretical or opaque
xeophon · x · 2026-08-16
The author expressed disappointment with the current state of AI safety work. They noted that most efforts are either purely theoretical without backing hypotheses or are closed off, providing only high-level, tainted summaries without substantial verification or transparency.
More from Safety
- Anthropic CEO responds to debate: Rejects the idea that regulation equals monopoly — basedjensen · 2026-08-16
- AI scams exploit holiday photos: tools pinpoint location for phishing — nordicinst · 2026-08-16
- Users Skeptical That Claude's Watermarking Won't Degrade Output Quality — AaronBergman18 · 2026-08-16
- Gavin Baker: Compute shortage buys civilization time — dr_alphalyrae · 2026-08-16
- AISI chief scientist departs to launch new nonprofit AI alignment research org — geoffreyirving · 2026-08-16
- Debate Erupts Over Anthropic Watermarking: Is It Technical Overreach or Plagiarism Prevention? — repligate · 2026-08-16