Enforcement Comes Into Focus in Calif. Digital Discrimination Bill

AB-2239 would codify in state law the FCC’s definition of digital discrimination. Republicans opposed the bill -- and some Democrats voiced reservations -- during a Communications Committee hearing earlier this month (see 2404110017). As cleared by that committee, the bill didn’t explain how enforcement would work, raising concerns for ISPs. The Judiciary panel addressed that void through a Tuesday amendment, which said that local public prosecutors and the California attorney general would enforce the prohibition on digital discrimination. Enforcers could seek injunctive relief rather than monetary damages, it said. Also, the amendment removed income level as a basis for liability.

"Seeking to strike a balance between holding broadband providers accountable while not opening a Pandora’s box of potential litigation," the amended bill "would utilize injunctive relief as the primary means of enforcing this measure,” a Judiciary Committee summary said. "Injunctive relief pursued by local prosecutors recognizes that broadband deployment requires significant coordination with local officials ... [A] local prosecutor is only likely to seek injunctive relief when a broadband company deviates from the previously approved plans in a discriminatory manner." The lack of monetary penalties obviates the need to define a safe harbor for companies to correct possibly discriminatory activities, it said.

The bill is "an incredibly crucial step towards promoting equity and fairness” since "even despite our best intents," many still lack equal access to essential broadband services, said Bonta. Assemblymember Rebecca Bauer-Kahan (D) supported the bill, which she said would be consistent with federal law. “We need to make sure that our state dollars are being equally used to provide access to a service now that has become critical to education, to employment, to the everyday things that people need to succeed in our communities,” she said. “Having an additional enforcement means to ensure nondiscrimination is only a good thing.”

Digital inclusion advocates lined up in support of the bill. EveryoneOn, a nonprofit, particularly likes the bill’s focus on “outcomes rather than intent,” testified Oscar Magana, senior programs manager-Los Angeles. But telecom industry groups including USTelecom, CTIA, the Wireless Infrastructure Association, the California Broadband & Video Association (CalBroadband) and the California Communications Association continued to oppose the bill.

CalBroadband is glad the committee removed income level as a basis for liability, said Amanda Gualderama, CalBroadband director-legislative and regulatory advocacy. However, the group fears AB-2239 “is still inconsistent with the Supreme Court precedent stating that a disparate impact standard must encompass any legitimate business decision, not just those based on feasibility,” she said. Also, the bill shouldn’t cover non-deployment factors like service quality, network management, customer service and contractual terms, Gualderama said. Plus, the group is concerned the “actual legal risk is much broader” than what the committee analysis predicts, she said. Finally, Gualderama asked why the California bill doesn’t presume like the FCC does that awarded projects in the broadband, equity, access and deployment program comply with non-discrimination rules. That could “lead to unintended consequences of discouraging participation in the federally funded grant programs,” she said. The Appropriations Committee will next weigh the bill.

Lawmakers Seek to Detect AI

The committee also advanced bills on election deepfakes (AB-2655), AI watermarking (AB-3211) and automated decision-making technology (AB-2930) to the Appropriations Committee. AB-2655 would require large online platforms to block posting and sending of election-related deepfakes and AI-generated content during election season. The bill would allow political content to post such content about themselves with a disclaimer. AB-3211 would require online platforms to label AI-generated content as fake. Meanwhile, AB-2930 would set rules for automated decision-making, including a ban on using the technology in a way that results in algorithmic discrimination.

Assemblymember Marc Berman (D) “can’t promise” his deepfakes bill “will survive a First Amendment or Section 230 court challenge," but "doing nothing isn't an option," he said. Berman noted that the proposal merely requires platforms to use the best available tools and only covers content the platform knows about. However, Bauer-Kahan said the bill could be more narrowly tailored to keep from running afoul of the First Amendment. Also, she said the definition of covered content seems overly broad, going beyond disinformation to include political speech. Berman said he’d keep working on the bill, which he termed a “work in progress.”

It would be hard for tech companies to comply with either the deepfake or watermarking bills, said witnesses for TechNet and the Computer & Communications Industry Association (CCIA). Platforms lack tools to definitively detect AI-generated content, testified CCIA State Policy Director Khara Boender: The proposed law could lead to over-blocking or over-labeling of content.

The Senate Judiciary Committee planned later Tuesday to consider multiple social-media bills and an anti-price-fixing bill (SB-1154) that would ban using algorithms to set or recommend prices using nonpublic competitor data. One social media bill (SB-1228) would require platforms to verify the identities of influential users and label them as authenticated or unauthenticated. SB-976, supported by California AG Rob Bonta (D), would provide social media controls for parents (see 2401300049). SB-1444 would require large social media platforms to provide APIs to allow third-party safety software companies to transfer minors' data and apply parental controls. And SB-1504 would update a cyberbullying law that requires social platforms to have reporting mechanisms. This year’s bill would require platforms to respond to cyberbullying reports and disclose their final determinations, while allowing minors to sue and multiplying by 10 the existing $7,500 daily, per-violation penalty.

CCIA raised concerns with most of the Senate bills in a Tuesday press release. “Without clarity in the bill language, it would be challenging to discern which data would be prohibited for training price algorithms” under SB-1154, said a statement by Boender. Also, she said legislators should reconsider social media bills, including SB-1228, SB-1444 and SB-1504, “that fail to address underlying pervasive issues or would cause unintended consequences and burdensome requirements without clear benefit to internet users in California.”