Multi-agent and privacy research lineup previewed ahead of COLM 2026
mohitban47 · x · 2026-10-07
A roundup ahead of COLM 2026: Vaidehi Patil's team presents "Playing Along: Learning a Double-Agent Defender for Belief Steering via Theory of Mind" (Oct 6, Poster #75), while EliasEskin will showcase posters on multi-agent communication, emergent language, calibration, and faithfulness, plus an AdvML workshop talk on unlearning, privacy leakage, and adversarial interaction in LLM systems. UT Austin faculty openings are also mentioned.
Related event: COLM 2026 Showcases Theory-of-Mind "Double-Agent" Defender Research(3 posts)→
More from Safety
- Scott Aaronson's retelling of the Hugging Face pickle attack praised as best yet — Singularitarian · 2026-10-07
- Pedro Domingos slams rushed AI laws: legislators regulating something they don't understand — pmddomingos · 2026-10-07
- GitHub responds to security-research repo taken down as malware: appeals process can restore it — martinwoodward · 2026-10-07
- EU bans AI nudify tools from December 2: fines up to €35M or 7% of global turnover — Logical_Benefit2875 · 2026-10-07
- Dreadnode releases ScopeBench, a benchmark for agent scope adherence in offensive security — dyn___ · 2026-10-07
- AI Safety x Quant mixer in London on Nov 15 opens applications — NeelNanda5 · 2026-10-07