Our recommendations aim to help governments, companies, and creators support a healthy, rights-respecting, and robust online speech environment. Hope you can check it out! cdt.org/insights/arc...

Political speech is indispensable and worth fighting for. At the same time, we all deserve an information ecosystem that’s transparent and empowers us to know who’s working (and paying) to win our votes and attention.

In "Architects of Online Influence," @isabelalinzer.bsky.social and I explore how political influencers shape online discourse — and how the financial and regulatory incentives surrounding them shape our democracy.

I think so.

The addition of the digital forgery crime that reintroduces an element regarding consent creates a lot more ambiguity about what was intended.

All of which is to say the absence of a consent-related element paired with a consent "rule of construction" is a remnant of the 2023 SHIELD Act. On its own, that bill was understandable enough, I think.

TAKE IT DOWN maintains the latter construction, with a privacy, but not a consent, related element for real depictions. But then TAKE IT DOWN creates a crime for digital forgeries with elements that are the exact opposite: directly addressing consent, but not privacy.

That appears to have shifted in 2023, removing the element addressing consent while leaving the element on privacy expectations, paired with language similar to what's in TAKE IT DOWN specifying consent to creation =/= consent to distribution www.congress.gov/bill/118th-c...

TAKE IT DOWN is basically the SHIELD Act, expanded to cover AI depictions, plus the new notice and take-down system. SHIELD has been introduced a few times over the years. When introduced in 2019, it had both lack of consent + privacy-related elements www.congress.gov/bill/116th-c...

As the industry works to come into compliance with the TAKE IT DOWN Act, they should look to our recommendations to build notice-and-takedown systems that are easy-to-use, victim-centered, and effective. cdt.org/wp-content/u...

In our report, we examined reporting mechanisms across 8 diverse platforms and identified troubling shortcomings - including inconsistent policy language and structure, insufficient incorporation of AI-generated NDII, barriers to reporting, and limited transparency and support.

When victims encounter NDII, their first response is often to seek the removal of that content to prevent its spread. That requires easy-to-use, accessible, and effective notice-and-takedown mechanisms for people in acute distress.

The non-consensual distribution of intimate imagery (NDII) — both real and AI-generated content — is a growing & deeply harmful form of image-based sexual abuse. With the proliferation of generative AI, victims face escalating risks and endure potential exposure even after content is taken down.

Careful consideration of the First Amendment stakes will ensure that consumers are appropriately protected from harms without opening the door to content- or viewpoint-based restrictions on the content we read and the information we seek.