Witnesses tell House panel to federalize protections against AI ‘deep fakes’ and require licensing for training data

Witnesses before a House subcommittee argued for a federal response to AI‑generated digital replicas and questioned whether current copyright and state laws are sufficient.

Sha Aston (introduced in the record as Mr. Aston), who identified himself as president of the performers union on the panel, described the emotional and reputational harm when a convincing deepfake appears online. "For the first time in human history, the experiences we've always shared are no longer secure," he said, urging Congress to protect consent, identity and livelihoods.

Dr. Bamma Viswanathan (academic witness) said digital replicas and unauthorized training on copyrighted works demand a federal solution. She recommended three pillars: judicial site‑blocking for clear offshore piracy cases, a federal property‑style right protecting identity and likeness, and market‑based licensing for works used to train large language models. "We have human rights and those rights deserve to be federalized," she told lawmakers.

Trade groups and technologists repeatedly warned about drafting details. Christopher Moore of the Software & Information Industry Association said his membership is split on the No Fakes Act and that key fixes are needed (preemption language, counter‑notice safeguards, intent requirements and careful scope to avoid chilling lawful tools). "If Congress moves forward, four guardrails should be non‑negotiable," he said, listing judicial process, targeting, safe harbors and flexible technical approaches.

Lawmakers pressed witnesses on fair use and recent litigation. Rep. Raskin and others referenced pending and recent cases (described in testimony as Barts v. Anthropic and a Northern District decision) to illustrate legal uncertainty about training datasets and fair use—witnesses said courts are still sorting those questions and encouraged licensing and market solutions to protect small creators.

On takedowns and speech protections, witnesses recommended procedural safeguards and counter‑notification mechanisms to reduce wrongful removals while preserving avenues for victims to remove authentic deepfakes. Several panelists emphasized exemptions for documentary, news, parody and other protected expression in drafted bills.

The chairman said committee staff will circulate legislative language and asked witnesses to respond to follow‑up questions; no bill was marked up during the hearing.