There was a time, not very long ago, when a photograph could settle an argument. If someone claimed the sky was green, a single picture of blue sky ended the discussion. That instinct—trusting our eyes, trusting a recorded moment—feels like a relic now. I remember my grandmother keeping a yellowed school photo in her wallet, pulling it out to prove a story about her childhood. She never questioned that the photo was true. It had to be true: it showed her sister laughing, a real morning in 1948. Today, that same photograph could be altered in seconds, or never have existed at all. The South China Morning Post opinion piece “Truth needs a treaty in this age of AI disinformation” captures this rupture with devastating clarity. The essay argues that artificial intelligence has not just made lying easier—it has industrialized deception, turning truth into a scarce resource while factories of synthetic content churn out fabricated realities around the clock. Deepfakes place words in mouths of world leaders; voice clones make parents hear children crying for help; fake news stories are tailored to each person’s fears, resentments, and hopes. Whatstrikes me in the article is its refusal to comfort us. It does not say “stay calm, technology will fix this.” Instead, it makes a case that feels foreign to our usual debates: that the world needs a treaty on truth, much like the treaties we created for nuclear weapons and chemical arms. Because AI disinformation does not respect national borders, no single government, platform, or editor can protect honest discourse alone. We need global agreements, shared red lines, mechanisms to hold everyone—including private companies and state actors—accountable. The article forced me to confront something uncomfortable: my generation was raised to believe that truth would win if we just had enough information. But information without agreed standards of evidence is not truth—it is just noise, and the loudest, most believable noise often wins.
The piece made me realize that we are not just facing a technological crisis; we are facing a spiritual one, a test of whether human societies can still agree on what is real. And when we can no longer agree on what is real, democracy, justice, and even simple conversation become impossible. That is why the treaty cannot be dismissed as unrealistic ambition. It is existential survival strategy.
.
The heart of the SCMP argument is an anatomy of how AI disinformation actually works, and it is more unsettling than any dystopian movie. The article does not dwell on killer robots or superintelligence; the danger is far more ordinary, far more human. AI does not need to be brilliant to destroy truth; it just needs to be cheap, fast, endless. A person can now generate thousands of fake articles overnight,, each one slightly different,, each one targeting a specific community’s fear: immigrants, vaccines, elections, climate change. The fabricated video of your local mayor accepting a bribe does not need to fool everyone—it only needs to reach enough skeptical people in time for the election. And once a lie has spread, correction is almost powerless; the first impression sticks like a burr to memory. The article describes how generative models have dismantled the trust anchors we relied on for centuries. We used to know that a voice belonged to a person because voices came from bodies. We used to know that video showed what happened because cameras were present. Now these anchors float free: seeing is no longer believing, hearing is no longer believing, and any amateur with a laptop can manufacture reality indistinguishable from documentary evidence. The piece also notes that older people, children, journalists, judges—all of us—are vulnerable in different ways. A retired veteran may be deceived by a fake video of a general endorsing a conspiracy; a teenager may lose faith in every news outlet after discovering one manipulated clip. The columnist’s central image stays with me: that we are all swimming in a polluted ocean of information, and a treaty is not about making the ocean perfectly clean—it is about stopping the factories dumping poison into it. That is a powerful reframe. It moves the problem from individual caution to collective responsibility. We have spent years telling people to be more skeptical, to fact-check, to pause before sharing. Those habits still matter, of course, but they put the burden on the drowning person rather than on the polluter. A treaty would begin to shift the burden where it belongs: on those who create and distribute disinformation at scale, and on the systems that profit from its spread.
The article then draws a brilliant analogy to past international cooperation, one that feels both hopeful and humbling. After chemical weapons killed thousands in World War I, nations did not simply ask soldiers to wear better masks; they eventually negotiated the Geneva Protocol, later the Chemical Weapons Convention, a framework declaring certain weapons illegal, invisible, illegitimate. Nuclear weapons were different: they could not be uninvented, so nations built a regime of treaties, inspections, non-proliferation commitments, and a fragile taboo that, so far, has prevented their use in war for decades. The SCMP author asks: why not apply the same logic to synthetic mediaand disinformation tools? The technology is advanced, scattered across countries, often hidden inside proprietary code, but that is exactly why we need public, global rules. The piece spells out what such a treaty might include: an international definition of prohibited deepfakes—those designed to incite violence, steal elections, or overthrow public trust; requirements for labeling AI-generated content robustly enough that manipulation cannot silently hide; shared databases of known disinformation campaigns; mutual assistance mechanisms so threatened countries can trace bot networks across borders; and penalties powerful enough to make rogue actors think twice. It acknowledges the obstacles honestly: authoritarian states may never sign; tech companies will lobby fiercely; enforcement will be imperfect; verification will be hard when algorithms are trade secrets. And yet, the article insists, imperfect treaties are better than no treaty. The Geneva conventions did not end war, but they changed what soldiers would openly do, and they made certain atrocities difficult to defend. A truth treaty would not eliminate lies, but it could build what the author calls “a moral horizon”—a line in the sand saying here, at least, digital fabrication to destabilize shared life is unacceptable. That line matters psychologically too. Without it, we gradually normalize constant deception, and we lose the ability to be outraged by the worst abuses. Once we normalize a world where anything can be faked, we are more likely to accept corruption, extremism, and the slow death of civic trust as inevitable. A treaty is a declaration that we still believe something is worth protecting.
What makes this opinion piece special, though, is that it refuses to let governments off the hook by pretending treaties alone will save us. The essay is deeply humanized in its final call: it reminds us that laws cannot replace relationships, institutions cannot replace courage, and international declarations cannot replace the small, daily acts of honest attention that keep a culture alive. The columnist writes almost tenderly about ordinary people—teachers, nurses, neighbours—who will never attend a diplomatic conference but who will determine whether truth survives. When a mother questions a video of her child in danger before she panics, when a voter asks where a story came from before she shares it, when a student cross-checks a claim against another source out of curiosity rather than fear, they are practicing treaty-making at the human level. The article calls for media literacy programs in every school, not as optional enrichment but as basic civil defense. It imagines community fact-checking networks where retired teachers and local librarians become digital guardians, helping neighbors step back from the algorithm’s rage. It asks for journalism that is slower, deeper, less eager to perform outrage, more committed to explaining what is known and what is not known. The piece also turns inward, urging us to examine our emotional vulnerability: why do we share disinformation? Usually not because we are stupid or evil, but because it confirms what we already want to believe, and it makes us feel part of a tribe. AI learns this vulnerability better than we know ourselves. It studies our clicks, our pauses, our fury, and it serves back our own reflection, polished to lethal certainty. A treaty can hold states accountable, but it cannot hold back the tide of rage inside us. That is why the author pairs international law with an older, humbler discipline: building the muscle of not-knowing, being comfortable with uncertainty, saying “I am not sure” in public without shame. There is something almost spiritual in this vision—truth as a practice of humility rather than a possession of the loudest voice.
The human dimension becomes even clearer when the article considers victims of disinformation. Too many policy debates stop at abstract terms—”information integrity,” “algorithmic transparency”—and forget that there is a person behind every lost reputation, every broken family, every violent mob. The piece tells of a local official accused by a deepfake of speaking a slur; her career collapses in week, before anyone verifies the video, and even after it is debunked, the residue follows her forever. Or imagine a hospital worker who loses her job because her voice was cloned maliciously, or a refugee family denied asylum because a fabricated video “showed” them committing a crime. These are not hypothetical edge cases; they are daily occurrences as AI disinformation tools spread. A treaty is ultimately about protecting people like these—people who cannot spend the hours to prove that the lie about them is false, who lack the platforms to correct the record, who suffer consequences that become politically explosive after it is too late. The article reminds us that disinformation is not a war machine only for states; it is also a weapon of the powerless, the desperate, the vengeful. That does not make it acceptable, but it demands that any solution be carefully designed not to simply silence anonymous speech or empower corporations to decide what is real. The author cautions that overbroad laws could be used by authoritarian regimes to jail journalists while claiming to fight “fake news.” The treaty must be precise: protect legitimate satire, protect whistleblowers, protect the awkward inconvenience of speech that is hard to verify but true. This complexity is not a reason to do nothing; it is a reason to do careful something. The essay captures this tension beautifully: “Truth does not need a ministry; it needs a community.” We must build institutions that are powerful enough to catch the poison but humble enough to know they are fallible. This is where the humanization lies: we are not building a world, but weaving a fabric, strand by strand, as we choose to pause, to ask, to listen, and to respect the dignity of evidence.
In its closing movement, the SCMP piece does something rare: it offers hope without dishonesty. It does not pretend that AI can be uninvented or that a treaty will magically restore the golden age of shared facts. The author acknowledges that this fight will be long, grinding, and frequently discouraging. There will be agreements betrayed, technologies outpacing rules, new deepfake methods invented the day after inspections begin. And yet, the essay insists, we are not doomed. History has shown that human societies can decide, collectively, that some things are too dangerous to leave unregulated. We did this with nuclear weapons, with acid rain, with ozone depletion—where scientists warned, citizens organized, diplomats negotiated, and slowly, painfully, institutions rose to meet the challenge. It may seem naïve to imagine a world treaty on truth, but it is far more naïve to imagine that unconstrained AI will nicely solve itself through goodwill. The columnist calls the moment we are living through “an inflection point where civilization either learns to govern its most powerful tools or becomes governed by them.” That phrase sizes in me. The greatest danger, maybe, is not that AI will destroy us spectacularly but that we will lose faith in one another one ordinary disagreement at a time. A treaty is, finally, a symbol of that faith. It says: despite our differences, despite national rivalries and corporate greed, we still believe that there is a shared reality worth preserving, a common world that we can investigate together with honest tools. The article ends with a call that feels more like a psalm than a policy paper: build the treaties, yes, but also build the table around which neighbors can talk; teach the children to verify; reward the journalist who corrects rather than defends; honor the citizen who admits she was wrong. Truth is not the absence of lies but the presence of care. In this age of AI disinformation, care is the only force heavy enough to anchor us. I close this reflection feeling enlarged—not comforted, exactly, but steadied. The SCMP opinion did not answer every question, but it gave us a direction. A treaty may be impossible today, but so was abolishing slavery, so was women’ssuffrage, so was the Montreal Protocol. The impossible becomes possible when enough people decide that the alternative is unliveable. Truth needs a treaty;and we, ordinary humans, are the ones who must demand it, one honest moment at a time.