With doomsday AI scenarios dominating the news, lawmakers are rightly concerned about reports concerning security breaches at major US AI labs, such as the OpenAI–Hugging Face incident and the many others reported in its aftermath. As they consider potentially regulating frontier AI, they should focus any new legislation on the immediate, demonstrated risks from those incidents. 

Post-incident reports show that the Hugging Face incident could have been mitigated or prevented by following longstanding cybersecurity best practices, like stronger sandboxing and monitoring. Any new legislation should focus on closing gaps in existing law to prevent AI companies from taking unreasonable risks with the public's security.

DONATE TO EFF 

ensure ai serves people, not power

When an AI developer or deployer runs a test or a task that has a high likelihood of causing harm to third parties—for instance, by breaking into someone else's computers—there should be clear minimum safety requirements. Such tests should run in a properly sandboxed test environment, disconnected from other systems, and be monitored and logged. Following these fundamental best practices would have prevented or substantially mitigated all of the incidents at AI labs that we currently know about.

That said, any proposal must be flexible enough to evolve with changing technology. Minimum safety requirements specific only to current AI technologies are likely to become obsolete; legal standards tied to well-established cybersecurity best practices are far more likely to stand the test of time. Tying any new mandates to evidence-backed security protocols also protects the public without impeding future AI development.

Strong legislation should also mandate and fund independent third-party investigations into any serious security incidents that may occur during AI labs’ tests of new tools, and make reports of these investigations available to the public. This important transparency measure would go a long way toward providing public oversight of the industry.

As with any technology regulation, those targeting cybersecurity practices at AI labs must be careful, precise, and practical.

Update: The Senate Commerce Committee voted to advance this bill on August 5, 2026. EFF continues to oppose the bill, which still needs approval from the full Senate. 

The Senate Commerce Committee is poised to consider the Youth AI Privacy Act, a bill that would require AI companies to create kids-only privacy rules and implement so-called “safe design features,” which would—like three other bills under consideration this week—require more data collection and make it harder for people to access lawful speech online. 

While the bill is narrower than some other proposed chatbot bills, it still has massive data security implications because it protects information for only certain users. This creates a problem we’ve cited many times before: if a bill requires that online services offer protections to minor users, the services will respond by imposing age gates to know which users should receive them. A better approach would be to offer the same privacy protections to all users. That way, we would avoid the services having to collect data on everyone to know a users’ age.

This bill also contains a problematic and vague provision that expressly allows AI companies to collect a known minor’s personal data for the purpose of testing, identifying, and addressing "harm to users”—without being clear on what exactly that means. Either way, services will need to collect even more information from young people, who are already targets of data theft and identity fraud. The Youth AI Privacy Act will give young people less privacy, not more. 

The Youth AI Privacy Act does include some positive privacy provisions around prohibiting the processing of personal information, like limiting what companies can do with people’s chat logs, including training, profiling, and disclosing them to other companies for training. But a general privacy bill must set these limits for everyone, not just minors. 

Mandating Design is Regulating Speech

The bill also requires the use of “safe design features,” which would restrict how online services providers design their systems and would deny teenagers the ability to use features like push alerts and notifications.  

We have seen this same type of restriction, sometimes called “age appropriate design code” in several states, including in California, Texas, and Arkansas. Unfortunately, these restrictions run into constitutional problems. In fact, federal courts have largely blocked these laws from going into effect because they likely violate the First Amendment rights of all internet users and the online services they regulate. Specifically, these laws interfere with internet users’ First Amendment rights to either speak or access speech online, and they also violate the rights on online services to decide how they will present information on their sites. 

Similarly, the Supreme Court has repeatedly ruled that “minors are entitled to a significant measure of First Amendment protection.” This does not mean that parents or guardians can’t set their own rules for their families—they can and they should, based on the needs and circumstances of the individual teenagers. But it does mean that Congress cannot adopt a “one size fits all” regulation that sets a restrictive government default that affects the First Amendment rights of all internet users, including teenagers.