REDDIT · REDDIT// 4h agoNEWS

Kolsetu eyes LLM 'absorption' privacy risk

Kolsetu addresses the "absorption" problem where LLMs store data residue in model weights, complicating GDPR compliance and the "right to be forgotten." The startup advocates for privacy-by-design AI agents that use deterministic playbooks to minimize risk in regulated industries.

// ANALYSIS

Kolsetu is highlighting a fundamental technical friction between LLM architecture and global privacy regulations.

–LLMs do not store data in discrete files but "absorb" it into billions of parameters, making traditional data deletion nearly impossible without retraining.
–Deterministic playbooks are positioned as a safer alternative to free-form generative outputs for industries like healthcare and finance.
–Their approach emphasizes regional data residency and strict input filtering as a bridge for enterprises currently blocked by compliance concerns.

// TAGS

kolsetullmsafetyethicsregulationagentgdpr

DISCOVERED

4h ago

2026-04-23

PUBLISHED

5h ago

2026-04-23

RELEVANCE

7/ 10

AUTHOR

EdikTheFurry