A gap Argentina's own law left open
In October 2023, Argentina enacted Ley Olimpia — Law 27.736 — which amended the country's comprehensive gender-violence statute (Law 26.485) to formally recognize "digital violence" as a form of gender-based violence, covering non-consensual distribution of intimate content, cyberstalking, doxing, and harassment, and empowering courts to order platforms to remove harmful material (Argentina.gob.ar; Boletín Oficial, Ley 27.736). What it does not do is name AI-generated or synthetic content as such — it was written for "real or edited" material, not images that never existed until a model produced them. Separately, deputies Eugenia Alianiello and Guillermo Snopek introduced a bill in December 2024 to add specific Penal Code articles for sexual deepfakes, arguing plainly that the code "does not specifically contemplate the creation, distribution or use of deepfakes with sexual content," leaving victims "without solid and effective legal tools" (Infobae). That bill has not passed.
Into this gap steps a different instrument. National deputy Carolina Basualdo (Provincias Unidas) has drafted a bill to create a labeling regime for digital content "generated, modified, or synthesized" using AI, and on June 18, 2026 she convened a public roundtable at the Córdoba Legislature with 17 academic, civil-society, and professional institutions — including the digital-rights group Fundeps — to workshop the draft before it heads to the national Chamber of Deputies (El Diario de Carlos Paz). Fundeps, recounting its participation, cited UN Women data for Latin America and the Caribbean that 96% of deepfake videos online consist of non-consensual sexual content, with women overwhelmingly the victims — the figure anchoring the bill's framing as a gender-violence measure, not just an AI-transparency one (Fundeps).
The case for labeling, stated fairly
The strongest version of this argument doesn't require exaggerating the harm — the 96% figure, while drawn from an older global dataset later corroborated by regional UN Women research, has held up for years, and Argentina's own prosecutors have had to improvise using injury statutes and child-protection law never written for synthetic media. A labeling mandate is also a comparatively light-touch tool: it doesn't ban AI content creation or compel platforms to pre-screen uploads, and it tracks disclosure obligations already moving through comparable regimes, such as the EU AI Act's Article 50 transparency requirements for synthetic and manipulated content. Applied to political and journalistic content, provenance labeling can plausibly reduce confusion without suppressing speech — a genuinely proportionate response to the disinformation half of the bill's mandate, which also touches electoral integrity and the labor rights of cultural workers whose likenesses get synthesized without consent.
Where the bill's logic breaks down
Fundeps says it "welcomes these multi-stakeholder dialogue spaces" and is contributing "evidence and policy recommendations" to keep AI policy grounded in human rights and gender perspectives.
But labeling and non-consensual sexual imagery are solving two different problems, and conflating them is where the bill's gender-violence framing runs into trouble. A label discloses that content is AI-generated; it does not stop that content from being made, uploaded to a foreign server, or circulated in the private groups that actually drive this harm. Nobody distributing non-consensual deepfake pornography is going to voluntarily tag it "AI-generated" — the entire point of the abuse is concealment, not disclosure. The tool that actually addresses the harm Basualdo's bill invokes is the takedown authority Ley Olimpia already created in 2023; the open question is whether Argentine courts and platforms are using it, not whether a new labeling statute exists on the books.
- Enforcement, not disclosure, is the binding constraint. Argentina's deepfake-abuse problem is a jurisdiction and platform-cooperation problem — most originating platforms sit outside Argentine courts' reach — that a domestic labeling statute cannot solve.
- Definitional scope matters enormously. "Generated, modified, or synthesized" is broad enough to sweep in routine AI-assisted editing tools used by Argentina's own growing creative-tech sector; without a tight carve-out for non-consensual sexual content specifically, the bill risks becoming a general content-labeling mandate justified by a harm it doesn't actually target.
- The draft is still pre-legislative. As of the June 18 roundtable, this is a stakeholder-workshopped draft, not an introduced bill — its final scope, penalties, and enforcement mechanism remain unwritten, which is exactly the moment to narrow it before it reaches the floor.
The proportionate path
The roundtable process itself — bringing in Fundeps, universities, and bar associations before drafting — is the right instinct, and it's worth crediting Basualdo for building the bill in public rather than around it. But the legislature would do more for the women Fundeps' data describes by funding enforcement of the takedown and evidence-preservation powers Ley Olimpia already grants, and by advancing the Alianiello-Snopek Penal Code amendments that directly criminalize non-consensual sexual synthesis, than by adding a disclosure requirement whose main practical effect will fall on legitimate AI use. Argentina doesn't need a broader definition of what must be labeled; it needs its existing digital-violence law to actually get used against the platforms and networks distributing this content today.