The Washington PostDemocracy Dies in Darkness

The Technology 202: As senators zero in on deepfakes, some experts fear their focus is misplaced

with Aaron Schaffer

A bipartisan group of lawmakers is moving ahead with legislation to combat deepfakes by launching a new federal task force — but experts question whether politicians and the public have been too focused on the nascent tech and not enough on cruder types of misleading content that constitutes a more immediate threat.

On Wednesday, a Senate panel unanimously advanced legislation to create a task force in the Department of Homeland Security charged with countering deepfakes — images manipulated through artificial intelligence used to fabricate events, such as faking that a politician or public figure said something they didn’t.

Officials in Washington have fretted for years about the threat posed by deepfakes, which they fear could sow discord during pivotal national moments, like an election, and wreak havoc on U.S. democracy.

It’s a rare area where lawmakers have found agreement in the battle against misleading online content, which has otherwise become highly politicized.

Tech leaders hailed the Senate Homeland Security and Governmental Affairs Committee for taking up the bill, the Deepfake Task Force Act, saying it will bolster government and industry efforts to identify whether content is authentic and verify its origins. The task force would comprise experts from government, industry, academia and civil society. 

“Creating a task force is a really smart way of starting a process before the problem becomes one that might require a sharp regulatory reaction,” said Aaron Cooper, vice president of global policy at BSA | The Software Alliance, a tech trade group that counts Oracle, Amazon Web Services and IBM as members. 

(Amazon founder Jeff Bezos owns The Washington Post.)

But as awareness about deepfakes has surged in recent years, experts have questioned whether the risk they pose has been overblown by politicians, the press and others.

“We’re going to be so focused on deepfakes … while spicy memes and authentic engagement online tear apart the shared set of facts that democracy depends on,” said Graham Brookie, director of the Atlantic Council’s Digital Forensic Research Lab, which studies misinformation.

Researchers are quick to point out that a vast majority of the misinformation and disinformation that they encounter online is far less sophisticated than a deepfake, which channels AI to create counterfeits that can be nearly indistinguishable to the human eye. Instead, it’s largely bogus memes, fake quotations, amateurishly altered videos and other crude forgeries. They also note that garden-variety forms of misinformation were the most prevalent during the 2016 election, which prompted a reckoning in the Bay and the Beltway over misleading posts online.

For example, Facebook was flooded in May 2019 by versions of video of House Speaker Nancy Pelosi (D-Calif.) that was altered to make her appear to be drunk and slurring her words. It wasn’t a deepfake but rather just simplistically slowed. 

Since 2016, policymakers in Washington have significantly dialed up their scrutiny of misinformation, holding numerous hearings on hoaxes and scams, hauling in social media executives to testify about how they’re handling them and peppering tech companies with letters about instances of misleading posts going viral. 

Despite the rise in attention, lawmakers have been unable to pass almost any legislation aimed at curtailing misinformation and disinformation online. That’s largely because the debate about what counts as misleading content has become deeply politicized.

So when senators announced plans to mark up and advance bipartisan legislation that zeroed in specifically on deepfakes, it raised some eyebrows online.

Brookie, who served on the National Security Council during the Obama administration, said widespread fear of deepfakes is “premature” because the technology hasn’t advanced enough to be both accessible and ubiquitous for those looking to stir up discontent online.

“The technology will get there at some point. … Deepfakes will become cost-effective at some point. That point is not now,” he said.

But there’s more to the bill than meets the eye. Despite its name, Brookie said, it actually could help combat other types of misinformation as well — not just deepfakes.

The legislation charges the task force with developing strategies to authenticate content, find its origin and trace how it’s changed through a chain of custody, a process known as digital provenance. “That’s a good trend line,” Brookie said.

Sen. Rob Portman of Ohio, the panel’s top Republican and the bill’s lead sponsor, said that will enable the task force to target both sophisticated and unsophisticated forms of misinformation, sometimes referred to as “cheapfakes.”

“By allowing people to track changes in digital content and authenticate that content as it moves around the Internet, digital provenance can work just as well to stop cheapfakes as it does deepfakes,” he told The Technology 202. 

Portman said the approach can “help combat not just deepfakes but related forms of mis- and disinformation, too.”

It’s also no small feat for lawmakers to agree on any legislation related to misleading content, given how polarized some of that debate has become, BSA’s Cooper said.

“It shows that while there are going to be aspects of all of these issues that can become partisan, there are also bipartisan approaches and members of Congress that want to find bipartisan approaches to many of the tech policy issues,” he said.

Our top tabs

The FTC blasted Facebook’s justification for shutting down access to researchers. 

A top Federal Trade Commission official dismissed Facebook’s justification for cutting off researchers’ access to data as “inaccurate,” I reported yesterday. Facebook said it disabled the accounts of researchers at the New York University Ad Observatory to comply with an agreement with the FTC. 

Samuel Levine, acting director for the Bureau of Consumer Protection, wrote in a letter to Facebook CEO Mark Zuckerberg that he was “disappointed” in the social media giant’s “inaccurate” justification for the move. Facebook did not respond to a request for comment.  

Apple will scan iPhones for child pornography, raising concerns about privacy. 

Apple will scan images stored on iPhones to identify the pictures, Reed Albergotti reports, as well as text messages for explicit content. There is less than a one-in-a-trillion chance of incorrectly flagging accounts, according to the company. 

Some privacy advocates are critical of the plans. “It’s impossible to build a client-side scanning system that can only be used for sexually explicit images sent or received by children,” the Electronic Frontier Foundation said. “As a consequence, even a well-intentioned effort to build such a system will break key promises of the messenger’s encryption itself and open the door to broader abuses.” Matthew D. Green, an associate professor at the Johns Hopkins Information Security Institute, argued the technology sends a message to repressive governments that “it is safe to build systems that scan users’ phones for prohibited content.” 

Apple will launch the program in its upcoming operating system. The company also told TechCrunch that it initially plans to roll it out in the United States, where Apple has long resisted calls by law enforcement to weaken its encryption technology. Other major technology companies, like Google, Facebook and Microsoft, also have systems to detect child pornography. 

The Senate is inching closer to advancing a major effort to make broadband more affordable. 

The roughly $1 trillion infrastructure proposal that the Senate is considering would amount to the largest ever one-time U.S. government investment in broadband, Tony Romm and Cat Zakrzewski report. Democrats and Republicans have expressed their reservations about some components of the measure but have largely agreed on the state of broadband in the bill. 

Final passage of the legislation is expected later this weekend. But it may still fall short of President Biden’s goal of ensuring that every American has access to high-speed Internet. 

“On the whole, this bill is going to help connect a whole lot of people,” according to Jenna Leventoff, senior policy counsel at consumer advocacy group Public Knowledge. “But there’s still a lot of work to do to make sure that we can fully close the digital divide.” 

Rant and rave

The FTC’s statement about Facebook and the NYU Ad Observatory didn’t go unnoticed on Twitter. Massachusetts Institute of Technology PhD student Willie Boag:

Former FTC chief technologist Ashkan Soltani and Ryan Calo, a professor at the University of Washington School of Law:

Software engineer James O’Leary had this amazing analogy:

Workforce report

Mapbox organizers have lost their union election vote (Protocol)


China’s lonely hearts reboot online romance with artificial intelligence (Alicia Chen and Lyric Li)

In 2001, a mysterious invention became a viral sensation — then went down in flames. I’ve always wondered: Was it all my fault? (Slate)

The latest TikTok trend is … Vietnamese music about the countryside? (Pitchfork)

Before you log off