Happy Tuesday! If you missed my interview with the FTC’s Alvaro Bedoya on Monday, check out a video from the event here. Send news tips to: cristiano.lima@washpost.com.
That includes laws in Arkansas and Utah requiring that parents sign off on their children’s access and a Florida proposal to ban those 16 and under from going on such sites altogether.
But a top federal enforcer active on kids’ safety issues said Monday that he opposes such limits, arguing they are unlikely to work and may run afoul of the Constitution.
“Meet a teenager, they will find a way to get around that,” Alvaro Bedoya, a Democratic member of the Federal Trade Commission, told me during an interview at the State of the Net conference in Washington. “I can say clearly I do not support that.”
Tech industry groups and digital rights advocates have expressed concern that to vet users’ ages and keep off teens, digital platforms will have to collect vast reams of information, raising thorny privacy concerns.
Furthermore, enforcing the laws could prove exceedingly difficult due to the technical challenges companies face in verifying users’ ages, both camps have argued.
Efforts to keep kids off social media through so-called “age-gating” laws have also faced mounting legal setbacks, with several state measures halted in federal court.
NetChoice, a tech trade group challenging many of the laws, has argued they violate the Constitution by blocking users from engaging in protected speech. The group counts Amazon, Google and Meta as members. (Amazon founder Jeff Bezos owns The Washington Post.)
The FTC’s Bedoya echoed those concerns, saying he has “a hard time seeing how” laws to block teens from accessing social media “would survive First Amendment scrutiny.”
Bedoya has called children’s online safety a top priority since joining the FTC in 2022 and has played a key role in the federal government’s response to the issue.
Since he joined the agency, the FTC has struck landmark settlements with video game developer Epic Games — maker of the Fortnite series that’s wildly popular with kids — over allegations the company deployed deceptive designs to trick consumers. In addition, the agency has proposed a sweeping plan to bar tech giant Meta from monetizing children’s data.
Bedoya is also a part of the Biden administration’s task force on children’s online safety, which is working to develop federal guidance for best practices companies can take to protect kids.
Bedoya largely declined to discuss ongoing enforcement actions, but he said one key issue for the agency is ensuring “kids and teens have to make the deliberate decision to connect with other people” so that “it isn’t imposed on them.” Another is scrutinizing the ways companies are trying to keep kids hooked on their offerings through their product designs, he said.
One way to stop kids from spending an unwanted or potentially harmful amount of time online, he said, is to limit how much companies can target them with ads.
Bedoya said he supports legislation banning targeted ads to kids, something lawmakers on Capitol Hill have proposed.
“It is a very compelling proposal to reduce the desire to keep children online in perpetuity, and targeted advertising is a key part of that,” he said.
Our top tabs
Threads starts testing ‘trending’ feature
Meta CEO Mark Zuckerberg announced Monday that Threads has begun testing a feature that shows the day’s trending topics, a step that could have implications for the app’s role in news and politics in a big election year, my colleague Will Oremus reports for The Technology 202.
The feature will display a selection of the topics getting the most engagement on Threads at any given time, as determined by the company’s machine-learning algorithms, a company spokesperson explained. A team of human “content specialists” will check them and remove topics that might violate Meta's policies.
Perhaps surprisingly, at a time when Meta is trying to play down political content on Threads and Instagram, the company said political topics will be eligible for inclusion in the trending list. Historically, trending topics have served as a popular window into a platform's discourse, sometimes sparking controversy when the topics are seen as problematic or biased in one way or another.
The feature is starting as a “small test” in the United States, Zuckerberg said in his post. “We’ll roll it out in more countries and languages once we get it tuned up.”
Helping the second class citizens of the AI boom
All the breakthroughs that we've come to expect from large language models like ChatGPT are virtually inaccessible in languages like Gujarati, Hausa, and Sinhala because most models have been built to fit around Chinese or English, my colleague Nitasha Tiku reports for The Technology 202.
“These two power languages are first class citizens and everything else has fallen by the wayside,” said Sara Hooker, vice president of research for the AI company Cohere and lead of C4AI, the company’s nonprofit research lab.
C4AI is helping to address that gap with the Aya project, an open source multilingual model covering 101 languages. More than 3,000 independent researchers from 119 countries contributed to the project, which showed a massive jump in quality for so-called “low resourced” languages that have little written, recorded, or digital material.
The Aya project focused on using native language speakers to develop smaller data sets for instruction fine-tuning. This is the type of data set that OpenAI used to make ChatGPT better at following instructions and responding to human preferences than its base model.
Glaring language gaps can create security flaws for all users, because models can be prompted to do things they shouldn’t in languages that it does poorly on, Hooker said. But the most profound impact from this class hierarchy could be distorting the types of data that will be created in the future, she said, comparing it to the internet, where English remains overrepresented because of the web’s early adopters.
Agency scanner
Inside the industry
Workforce report
Trending
Daybook
- Cyberscoop hosts the “Zero Trust Summit” on Thursday at 8 a.m.
- The House Science Committee holds a hearing, “Examining Federal Science Agency Actions to Secure the U.S. Science and Technology Enterprise,” Thursday at 10 a.m.
- The House Committee on Veterans’ Affairs holds a hearing, “Artificial Intelligence at VA: Exploring its Current State and Future Possibilities,” Thursday at 10 a.m.
- The House Financial Services Committee holds a hearing, “Crypto Crime in Context Part II: Examining Approaches to Combat Illicit Activity,” Thursday at 2 p.m.
Before you log off
Tell me Im wrong pic.twitter.com/cOL34vSxz7
— Kevin Ryan (@KevinRyanComedy) February 12, 2024
That’s all for today — thank you so much for joining us! Make sure to tell others to subscribe to The Technology 202 here. Get in touch with Cristiano (via email or social media) and Will (via email or social media) for tips, feedback or greetings!
