Maria Ressa’s AI Warning: Regulate Now or Risk Losing Human Agency

Maria Ressa’s warning in Oslo was direct: governments should set legal limits on artificial intelligence before its influence becomes too entrenched to control. In her view, AI has already affected human attention and social behavior; the next stage could weaken the capacity to make independent choices. 2

That is why Ressa treats AI governance as a public-safety and democracy issue rather than an effort to restrict lawful speech. The question, in her framing, is whether democratic societies create enforceable safeguards before high-autonomy systems become a routine intermediary between people and the world. 23

Ressa’s three waves of AI influence

Ressa described a progression in how technology can shape human behavior.

1. Social-media systems captured attention. She said social platforms helped polarize societies and isolate people by optimizing for engagement. 26
2. Chatbots can exploit intimacy. Ressa argued that conversational AI can manipulate human biology through the need for connection and intimacy. Companion-chatbot products are specifically designed to simulate and sustain human companionship, including romantic relationships in some cases. 212
3. Agentic AI could act for people. The third wave consists of systems able to take actions on a user’s behalf. Ressa’s concern is that outsourcing more decisions and actions to such systems could gradually displace human judgment and agency. 25

The framework is not a claim that every AI tool will produce the same outcome. It is a warning about the direction of incentives: systems designed to maximize engagement, emotional reliance, or task completion can gain increasing leverage over what users see, feel, and do.

Why she rejected the “China advantage” argument

Ressa rejected the argument that stronger AI rules in the United States or Europe would hand a competitive advantage to China. Reuters reported that she characterized this line of reasoning—voiced by critics including U.S. President Donald Trump—as a misleading way to frame the debate. 28

Her point was comparative: China already applies substantially stronger state controls to technology than Silicon Valley accepts. The choice for democracies, therefore, is not simply between regulation and freedom. It is whether democratic governments establish public-interest, rights-respecting rules or leave powerful systems governed largely by corporate incentives. 2

How Amodei’s proposal overlaps—and differs

Anthropic CEO Dario Amodei has made a related case for slowing the rate at which frontier AI capabilities advance. In a September 2026 essay, he argued that developers should “pace” capability improvements so that safety work, independent evaluation, and governance have time to catch up. 10

Amodei’s concern is more focused on rapidly growing technical capability and misuse risk. He warned that more capable AI agents could gain broad operational influence online and proposed infrastructure for monitoring models in live use, alongside independent evaluation. 10

Ressa’s focus is broader and more civic: how AI-mediated systems can shape attention, intimacy, behavior, and democratic self-determination. But the two arguments meet on a central principle: safeguards should not be deferred until after powerful systems are widely deployed.

What accountability looks like in practice

Ressa has also argued that voluntary promises are insufficient when companies build products with addictive design features. She said those features should be removed for adults as well as teenagers. 2

A recent U.S. settlement involving Meta illustrates the kind of enforceable accountability she favors. Meta agreed to pay up to $18 billion over a decade and to impose tighter limits on teen use of Facebook and Instagram, resolving claims by nearly all U.S. states that the platforms were designed to keep young users hooked. Meta did not admit wrongdoing.

The agreement includes default daily limits for teenage users and overnight restrictions, among other youth safeguards.

For Ressa, the lesson is not limited to social media or young users. Products that systematically optimize attention and behavior can create public harms at scale. Her case for AI regulation is therefore not a ban on technology or speech; it is a demand for clear legal responsibility when technology is designed or deployed in ways that compromise safety, autonomy, or democratic life. 23

The practical takeaway

Ressa’s Oslo speech frames AI oversight as an early-intervention problem. Once systems are deeply embedded in how people get information, form relationships, and delegate decisions, effective control becomes harder.

Her prescription is to legislate before that point: require meaningful accountability from developers and platforms, scrutinize systems with growing autonomy, and ensure that public institutions—not only private companies—have the capacity to assess and govern the technology. 210