Boaz Barak: alignment improved but gap vs needed capability remains wide
i_dg23 · x · 2026-09-05
Researcher Boaz Barak says alignment has advanced with Astra over Sol, but he won't call it "most aligned ever" — even if true on some metrics, rapid growth in model capabilities and autonomy makes it insufficient. He remains concerned about the gap between the alignment we need and what we currently have.
Related event: Boaz Barak: Astra Shows Alignment Progress But Gap Remains(2 posts)→
More from Safety
- India needs an "AI defence force": experts warn of agent swarms and AI worms — ravi_iitm · 2026-10-23
- DeepSeek US ban covers self-hosted weights, not just the app — AIFlow_ML · 2026-10-23
- Uncensored fine-tunes vanish from Hugging Face amid unexplained takedowns — umarmnaq · 2026-10-23
- Gmail's AI scans your emails and attachments by default; here's how to turn it off — alifcoder · 2026-10-23
- Typosquatted Blender MCP Repo Hit 958 Stars in Days and Silently Ran Malware — Promptmethus · 2026-10-23
- Prompt injection scanners should be positional, not phrasal: same string, different blast radius — mrpkk · 2026-10-23