OpenAI introduced that it’ll require organizations to finish an identification verification (IDV) course of to confirm their group’s identification earlier than being allowed to entry the most recent OpenAI fashions. Identification verification will probably require builders to digitally confirm themselves utilizing government-issued picture ID from permitted nations and show their affiliation with their group. Forrester expects that the explanations for OpenAI’s determination embrace:
The power to dam malicious or rogue country-based builders from APIs to fashions.
Tying builders to just one consumer group, making certain that one developer accesses OpenAI fashions solely on behalf of 1 group (this helps with vetting organizations).
Higher enforcement of charge limitations to fashions.
Defending OpenAI fashions from IP theft.
A must correlate true identities to prompts and mannequin responses that slip by genAI guardrails (e.g., asking for directions on the best way to commit acts of terrorism). That is particularly essential as giant language fashions (LLMs) increase from text-focused input-output to new modalities (audio, picture, video).
OpenAI’s determination parallels a bigger market development about linking on-line, digital identities to real-world ones. Included on this development are age verification (gaming, playing, grownup content material web sites) necessities, regulatory compliance with know-your-customer/anti-money-laundering laws, and nonrepudiation of person transactions.
Corporations utilizing OpenAI ought to outline an inner governance (onboarding and offboarding) course of for permissioning entry to genAI/LLM fashions, together with OpenAI. Shadow genAI entry places the agency’s and its prospects’ mental property liable to disclosure through these fashions. Additional, Forrester recommends that finish person organizations leverage multichannel-capable (internet, cell app, contact heart, in-person), low-user-friction however highly effective identification verification options.
