FCC's Final NPRM on AI and Robocalls Had Many Changes Over Draft

The FCC’s NPRM on AI and robocalls that commissioners approved Wednesday saw numerous changes from its draft version, beyond the addition of a notice of inquiry (see 2408070037). Incompas and the Cloud Communications Alliance asked the FCC to move parts to a NOI, citing the lack of specific proposals (see 2408050029). “The item itself is seemingly more of an investigation into the state of AI technologies rather than a series of specific proposals,” they said.

Commissioners agreed to move two areas of questions to an NOI, a preliminary step to an NPRM. Those questions deal with technological challenges and privacy. Also shifted to the NOI were questions about the National Institute of Standards and Technology’s AI Risk Management Framework. Many of the questions remain the same as in the draft.

“Should the Commission act to further the development and deployment of such technologies?” the FCC now asks. Questions remain about legal, technical and practical barriers “to wide-scale deployment and adoption.” But the final version now asks whether the commission has the statutory authority and technical expertise to address those barriers. The NOI section also asks about the timeline and cost for carriers to upgrade their networks.

The FCC also now inquires about the implications of secure telephone identity revisited (Stir) and signature-based handling of asserted information using tokens (Shaken) authentication standards. “To what extent do these technologies duplicate or complement STIR/SHAKEN and other caller ID authentication solutions?” the notice asks: “How do issues regarding IP interconnection across voice service networks impact the ability of providers to enable real-time monitoring of voice traffic using AI technologies?”

The FCC also changed a section on the definition of an AI-generated call. The final version notes that this is only a “proposed” definition. It adds a paragraph that was not included in the draft. “Alternatively, we seek comment on whether it is necessary to define 'AI-generated call' with specificity, given that the [Telephone Consumer Protection Act] expressly covers 'artificial or prerecorded voice,' and given that we have already determined that voice cloning and similar technologies qualify under that statutory phrase,” the new section says: “If we do not define an AI-generated call in this context, how would callers determine whether the disclosure obligations proposed below apply to the calls and texts messages that they are sending?”

A section on call disclosure contains more specificity now. It says disclosure is intended “to facilitate consumers’ ability to make an informed decision to manage unwanted calls” and that the disclosure should include “certain information that would enable the called party to identify the person or entity initiating the call.”

The FCC also added to the questions posed about consumer consent. The NPRM now asks whether for calls that already require prior express consent, “would it benefit consumers to require them to provide separate consent to receive AI-generated calls?” It notes that “in reliance on our prior express consent framework dating back several decades, many callers have already captured prior express consent to place autodialed and/or artificial or prerecorded voice calls in a manner that comports with the TCPA and our current rules.”

The NPRM seeks comment on “the potential benefits and drawbacks” of AI-specific disclosures. “Would consumers benefit from new disclosures that apply to ‘AI-generated calls,’ but not to ‘artificial or prerecorded voice’ calls outside the new definition?” the NPRM asks.

The NPRM was in Friday’s Daily Digest. Comment deadlines will come in a Federal Register notice. All five commissioners produced written statements.