Understanding Section 230: The Law That Shaped the Internet

The Digital Service Act (DSA) represents the EU's comprehensive internet regulation framework, serving as the new 'constitution' for online governance. Unlike outdated 2000s laws, the DSA addresses modern challenges including aggressive algorithms, violent content, and manipulative marketing. The core principle is that illegal offline activities must also be illegal online, extending legal protections to digital spaces. The law prohibits 'dark patterns' that deceive users, requires chronological feed options, and implements tiered regulation based on platform size. Large platforms with over 45 million users face strict scrutiny, while smaller platforms face lighter requirements. The DSA establishes external arbitration systems for disputes, with over 50 million decisions reversed and users winning 52% of cases. The law requires platforms to explain recommendation criteria, prohibit targeted advertising to minors, and maintain public advertising registries. Penalties reach 6% of global annual turnover, with the EU having already fined major platforms for violations.

The Digital Services Act (DSA) is an EU regulation establishing consumer protection mechanisms for online platforms. Users facing unjustified content blocking can file complaints through designated coordinators like the Czech Telecommunications Office. The regulation requires platforms to restore content if blocking was unjustified, creating a formal dispute resolution framework. However, the legislation contains both technical provisions for internet safety and ideological elements requiring national implementation, creating complex challenges for member states.

The EU Digital Services Act (DSA) is a comprehensive internet regulation regime passed in 2022 by the EU legislature, becoming applicable in 2023. The DSA requires very large online platforms (VLOPs) operating in the EU, such as X, Facebook, and YouTube, to censor broad and vague categories of online speech, including alleged misinformation. The DSA represents Europe's comprehensive approach to internet regulation, with the stated objective of ensuring everyone in the EU can enjoy online platform services safely while respecting fundamental rights including privacy, dignity, protection of minors, security, democracy, and freedom of speech.

The European Union has implemented two key digital regulations: the Digital Service Act (DSA) and the Digital Markets Act (DMA). The DSA, which entered into force in August 2023, holds 19 major platforms accountable for illegal content and harmful material. The DMA addresses anti-competitive practices by major tech companies (GAFAM). These regulations were designed to protect consumers, minors, and ensure fair competition, but their application to platforms like X has sparked debates about free speech and platform governance.

The Digital Services Act (DSA) is described as one of the most restrictive laws in the EU, with provisions that critics compare to Soviet-era directives from the NKVD. The speaker notes that EU leaders are called 'commissars' rather than 'ministers,' drawing parallels to Soviet bureaucratic structures. This legislation is presented as part of a broader pattern of authoritarian-style governance within EU institutions.

Social media algorithms that curate and recommend content represent editorial judgment by the platform. The Supreme Court has recognized that when platforms use algorithms to select and present third-party content in a curated manner, this constitutes the platform's own speech protected by the First Amendment. This means that algorithmic recommendations are treated as the platform's own expression rather than mere passive distribution. Consequently, Section 230 immunity does not protect platforms from liability for harms resulting from their own algorithmic curation decisions.

Section 230 protects platforms from liability for third-party content, including algorithmic curation, because any content organization method—including reverse chronological feeds—is fundamentally an editorial choice; removing these protections would collapse the internet into a fragmented hosting-only environment, harming small platforms and communities while empowering government censorship over content moderation decisions.

Section 230 of the Communications Decency Act grants immunity to internet platforms for user-generated content, distinguishing between hosting content and actively curating or recommending it; the Supreme Court's upcoming decision in Gonzalez v. Google could fundamentally reshape internet liability by determining whether algorithmic recommendations constitute editorial control that falls outside Section 230's protection, with potential consequences for free speech and platform operations.

In the Gonzalez v. Google case, the U.S. Supreme Court ruled that algorithmic content recommendations do not fall outside the scope of Section 230 of the Communications Decency Act, meaning platforms like YouTube can continue operating normally without facing publisher liability for content they recommend; the Court also unanimously dismissed similar claims against Twitter in the Taamneh case, establishing that web hosts cannot be considered aiding and abetting bad actors because they have equal ability to upload content to their sites.

Section 230 of the Communications Decency Act grants internet platforms featuring user-generated content immunity from being sued over content published by those users. In the case of a 10-year-old girl who killed herself participating in a Blackout Challenge on TikTok, an appeals court reversed a ruling that Section 230 protected TikTok's algorithm, arguing that the algorithm represents a unique expressive product that communicates to users through a curated stream of videos. Algorithms are mathematical equations of variables and weights that translate the multimodal model in someone's brain into a model that sits in code. This is similar to how an editor 50 years ago had an equation of variables and weights in their mind. The current moment is that computers are thinking actively for us, and we need to rewrite laws through the lens of the 21st century. An algorithm is supposed to give you more of what you want, while an editorial page has a specific point of view. If you're pro-Trump, you'll see more pro-Trump content; if you're pro-Harris, you'll see more Harris content. The platform isn't taking an editorial position on whether it supports one or the other - it's giving you more of what you're interacting with. This is fundamentally different from having an editorial point of view.

State-level regulations protecting children from harmful online content face constitutional challenges under First Amendment protections. California's online child safety law requires social media companies to implement age verification, content filtering, and privacy protections for young users. Tech industry organizations challenged the law, arguing it transforms companies into government censorship agents. Federal courts partially upheld the law while striking down certain provisions, demonstrating the ongoing tension between state regulatory authority and constitutional free speech protections in the digital age.

The Supreme Court declined to block Mississippi's new social media law requiring all users to verify their ages before accessing sites like Facebook, YouTube, and Instagram. Justice Brett Kavanaugh concurred that under current case law, the law is likely unconstitutional but noted the challenging party had not sufficiently demonstrated that harms to social media sites outweighed allowing the law to remain in effect. The law expands beyond Texas's recent age verification law by additionally prohibiting minors from using popular social media without parental consent and requiring platforms to prevent children from accessing harmful materials. The case was sent back to lower courts while remaining in force.

The Communications Decency Act was enacted in 1996, and the Supreme Court has never provided a substantive interpretation of the most significant language in the statute. The phrase 'a lawsuit cannot treat an internet platform like the publisher or speaker of someone else's information' has become one of the broadest immunities in law. This raises the question of whether reform should come from Congress or from the Supreme Court interpreting the statute's language.

Despite years of calls for federal oversight, Congress has failed to regulate tech giants effectively. This vacuum has enabled state-level legal action to emerge as the primary mechanism for holding companies accountable. Multiple state attorneys general have filed similar cases against Meta, YouTube, Snap, and TikTok. The legal system typically lags a decade behind technological developments, meaning courts are now addressing problems visible for over a decade. These cases may ultimately reach the Supreme Court, representing a fundamental shift in how technology regulation is pursued in the United States.

The Supreme Court refused to review a decision from Vermont's Supreme Court that allows the state to pursue claims against the social media company Meta. Vermont's attorney general is suing Instagram, claiming it violated state law intended to curb social media addiction, but Meta argues the state's courts don't have jurisdiction to review the case. The Supreme Court allowing the case to move forward could impact cases brought by other states. Meta and YouTube are both fighting hundreds of lawsuits brought by school districts and AGs across the country.

Section 230 functions as a meta-law—a law that determines who governs an emerging industry rather than how governance occurs. Enacted when the internet was still in its infancy, Section 230's consequences proved extremely difficult to correct once the industry matured. The very first meta-law on technology often becomes the last major law because it determines who has power to resist future corrections. Section 230 may be the 26 words that created the internet, but they also visited irrevocable harm to generations of children. For generative AI, Section 230 should not apply because AI outputs are the company's own creations, not third-party content. Federal preemption of state laws for AI would repeat Section 230's tragic history without corrective measures. Congress should not establish broad immunity for emerging industries it doesn't understand.

The speaker addresses whether Section 230's benefit of immunity should apply to situations where platforms promote harmful content to users, such as the case of a 10-year-old girl who died from a deadly Blackout Challenge that encouraged her to choke herself. The speaker argues that Section 230's immunity should apply very narrowly for two kinds of situations: one is when the platform is taking active steps to mitigate against harm, and the other is when the question is whether someone is being treated as though they are the speaker for someone else's speech. The speaker argues that the facts described would not fit either of these narrow views. The speaker also addresses how Section 230 has allowed activities such as online sex trafficking, exploitation, and pornography to exist and grow. The speaker argues that courts have taken what is a fairly clear text and turned it on its head, and they are being led to that point by litigants who are arguing for this massively broad immunity. The speaker emphasizes that Section 230 has been abused in this way, denying people the opportunity to get into court where they can show how these companies are in fact engaged in that activity. The speaker addresses how Section 230 has failed to protect children from harmful content online. The speaker argues that Section 230 was not intended and does not provide civil liability immunity for content created by generative AI, including child sexual abuse material. The speaker argues that Section 230 has not stood up to its original intent to protect children, and that the numbers of cases demonstrate categorically and quantifiably that Section 230 has failed to protect children from exploitation and exposure to obscene material. The speaker emphasizes that this is amplified by a lack of liability for these companies under Section 230. The speaker also addresses how Section 230 has impacted civil rights and the rise of generative AI, noting that these technologies have democratized the use of terrible tools and practices that used to be expensive and hard to operate. The speaker argues that one of the explicit targets for reform for Section 230 needs to be to spell out that for generative AI, they do not get the protections of this immunity.

Section 230, which absolved social media companies from liability for user-generated content, does not apply to generative AI according to OpenAI's CEO. For a very new technology, a new framework is needed. Companies bear responsibility for the tools they put out in the world, but tool users also bear responsibility. The question of how to structure liability between companies, users, and those who build on top of the technology is super important. The government's previous approach of getting out of the way of new industries may have led to harms that should not be repeated.

Senator Graham raises the question of whether Section 230 (which provides immunity for online platforms) should apply to generative AI. OpenAI's CEO acknowledges that Section 230 doesn't apply to generative AI and that developers should not be entitled to full immunity for harms caused by their products. This raises questions about what liability framework should apply to AI companies.

Section 230, which provides immunity for online platforms, does not apply to generative AI according to OpenAI's CEO. This means developers like OpenAI should not be entitled to full immunity for harms caused by their products. The lesson from Section 230 is that when government gets out of the way of new industries, problems can result (child exploitation, cyberbullying, online drug sales). The opposite approach is needed for AI—coming in on the front end to establish liability standards and precision regulation. Trust is the license to operate, and companies must build not undermine public trust.