The TAKE IT DOWN Act: First Federal Deepfake Law Explained | State of Surveillance

TL;DR: On May 19, 2025, President Trump signed the TAKE IT DOWN Act, the first federal law specifically targeting non-consensual intimate images and deepfakes. Sharing such content is now a federal crime: up to 2 years for adult victims, 3 years for minors. Platforms must remove reported content within 48 hours. The law takes effect May 2026. It passed 409-2 in the House. Meta, Google, and TikTok supported it. The EFF has concerns about free speech implications. Here's what you need to know.

What the Law Does

The "Tools to Address Known Exploitation by Immobilizing Technological Deepfakes on Websites and Networks Act" (TAKE IT DOWN) creates federal crimes and platform requirements for non-consensual intimate imagery ^[1].

It prohibits:

Publishing intimate images of anyone without their consent
Publishing deepfakes (AI-generated) intended to cause harm
Threatening to publish intimate images or deepfakes
Using such threats for intimidation, coercion, or extortion

It requires platforms to:

Create a notice-and-takedown process for reporting
Remove reported content within 48 hours
Make "reasonable efforts" to remove duplicates and reposts

This is the first federal law specifically addressing deepfakes. Previously, victims had to rely on a patchwork of state laws, if their state had any.

Criminal Penalties

Creating or sharing non-consensual intimate imagery is now a federal crime:

Offense	Maximum Prison	Fines
Adult victim	2 years	Federal fines
Minor victim	3 years	Federal fines
Threats to publish	2-3 years	Federal fines

FTC enforcement: The FTC can seek civil penalties up to $53,088 per violation, plus consumer redress and injunctive relief ^[2].

That's not theoretical. If someone creates a deepfake of you and posts it, they're now looking at federal prison time.

The 48-Hour Rule

Starting May 2026, platforms must respond to reports within 48 hours ^[3].

How the process works:

Victim submits report with electronic signature
Report includes statement of good faith belief content was non-consensual
Report provides information to locate the content
Platform has 48 hours to investigate and remove
Platform must try to remove duplicates/reposts

Platform protections: If a platform removes content in good faith based on a report, they're protected from liability, even if the content later turns out to be legal. This encourages platforms to err on the side of removal.

That's a significant shift. Previously, platforms could ignore reports indefinitely. Now there's a federal deadline.

Who Supported This

The bill had unusually broad support:

In Congress:

Passed House 409-2
Near-unanimous Senate vote
Introduced by Senator Ted Cruz (R-TX)
Co-sponsored by Senator Amy Klobuchar (D-MN)

Tech companies supporting it:

Meta
Google
TikTok
Over 100 organizations total

First Lady Melania Trump championed the bill as part of her "Be Best" anti-cyberbullying campaign. Trump invited her to co-sign the law ^[4].

When Meta, Google, and TikTok all support the same legislation, something unusual is happening. In this case: they want clear federal rules instead of 50 different state laws.

The Concerns

Civil liberties groups aren't celebrating. The EFF, Center for Democracy & Technology, and Freedom of the Press Foundation raised concerns ^[5]:

Vague language: The law targets "intimate images" but doesn't clearly exempt:

Satire and parody
Newsworthy content
Political speech
Non-public content stored on servers

Encryption risks: The requirement to remove content "stored on servers" could pressure providers to break end-to-end encryption to comply.

Abuse potential: The notice-and-takedown system could be weaponized by bad-faith actors to remove legitimate content by filing false reports.

Over-removal: Platform liability protections encourage removing first, asking questions later. Legitimate content could get caught in the dragnet.

These aren't hypothetical concerns. DMCA takedown systems are routinely abused. This law creates a similar system for intimate imagery.

Why This Law, Why Now

The explosion of AI-generated deepfakes forced Congress to act:

Grok generating non-consensual sexual imagery of anyone on X
AI CSAM increasing by "orders of magnitude" in 2025
Taylor Swift deepfakes going viral, sparking public outrage
School incidents with students creating deepfakes of classmates
48 states passing their own deepfake laws, creating regulatory chaos

The law is a response to AI tools making abuse trivially easy. Anyone with a smartphone can now create convincing fake intimate imagery of anyone else.

Congress didn't move fast because they suddenly cared about victims. They moved because the problem became impossible to ignore.

What This Means For You

If you're a victim:

Starting May 2026, you can file federal reports for removal
Platforms must respond within 48 hours
Creators face federal criminal charges
You can pursue civil remedies through the FTC

If you're on platforms:

Expect new reporting mechanisms rolling out
Content removal will likely increase
False reports may affect legitimate content

If you create content:

AI-generated intimate imagery of real people = federal crime
Even "satirical" deepfakes could face prosecution
The line between legal and illegal isn't clearly drawn

What It Doesn't Do

The TAKE IT DOWN Act has significant gaps:

Doesn't address AI tools themselves. Creating the tools is still legal
Doesn't require proactive detection. Platforms only act on reports
Doesn't cover all deepfakes. Only "intimate" imagery or content "intended to cause harm"
Doesn't address political deepfakes. Election misinformation falls under separate (weak) laws
Enforcement depends on victims reporting. Many won't know they've been targeted

Grok can still generate non-consensual imagery. This law just makes sharing it a crime and requires platforms to remove it when reported. The AI that creates the content remains unregulated.

The Bottom Line

The TAKE IT DOWN Act is the first federal law against deepfakes. It creates real criminal penalties and real platform requirements. Victims now have federal recourse.

But it's reactive, not preventive. It addresses symptoms while leaving the disease untreated. AI tools that generate abuse remain legal. Platforms only have to act when someone reports content, not before.

It's a start. Whether it's enough depends on how aggressively it's enforced and whether the civil liberties concerns prove justified.

The law takes effect May 2026. Platforms are already preparing. The question is whether victims will actually see protection, or just another system to navigate while their images spread.