Joshua Gans proposes licensing AI training ex post: compensate creators only when outputs substitute
joshgans · x · 2026-10-30
In "Permission to Learn: An Economic Approach to AI Copyright" (Network Law Review), U. of Toronto professor Joshua Gans argues that training licences cannot be negotiated in advance. He proposes letting training proceed and compensating creators only once model outputs actually substitute for their work — shifting enforcement from the training stage to output-stage substitution.
More from Safety
- Security researcher memes about the flood of RCEs and full chains — moyix · 2026-10-30
- DeepSeek app jailbroken in under a minute to spew RAT malware code — rez0__ · 2026-10-30
- AGI House dinner notes: Anthropic, Harvey AI execs meet policymakers — agihouse_org · 2026-10-30
- Agent verification failure log: the patch was real, but it fixed a different bug — AIGODSEND · 2026-10-30
- House Dems push resolution opposing taxpayer bailouts for failing AI companies — michalmalewicz · 2026-10-30
- Crypto dev eats his words on public Wi-Fi safety after eMudhra TLS surprise — matthew_d_green · 2026-10-30