What the Schloms–Monet episode reveals about AI and authorship’s future

Last May, on X, anonymous artist Schloms posted a cropped painting and claimed he had “just generated an image in Monet’s style.” He asked the crowd to explain why it fell short of the real thing. The post went viral. Within hours, the social jury tore it apart: no depth, incoherent strokes, off colors, “digital trash.” Then came the reveal: it was a real Monet hanging in Munich. When the truth surfaced, many deleted their takes. Schloms had watched, and he published the screenshots. A single label had triggered cognitive repulsion.

What’s striking is not only the bias but its target: the notion that AI drains art of its human center. We look for a hand, a breath, a decision in the marks. In AI we project fears of losing uniqueness, talent, authorship. Yet the words used—“confused, formless, soulless”—echo the barbs thrown at the Impressionists in the 19th century. Every visual revolution has first been read as a loss of humanity. The difference now is that AI shifts not only what can count as art, but who can count as an artist.

The implications are immediate. Technically, institutions are moving toward traceability standards: robust watermarks, machine-readable labels, provenance signals. The European Union’s AI Act, adopted on 13 March 2024 and published on 12 July 2024, mandates transparency for generative systems, including labels for synthetic media to mitigate deception and disinformation (source: EUR-Lex, AI Act). UNESCO’s 2021 Recommendation on the Ethics of AI urges traceability and accountability in cultural applications (source: UNESCO). The World Intellectual Property Organization’s 2019–2024 reports map the paradox: AI amplifies creative output while straining authorship and originality doctrines, pushing for copyright clarity (source: WIPO).

Legally, guardrails are tightening. The EU’s 2019/790 directive on digital copyright provides conditional TDM exceptions and pushes source transparency; in the U.S., the Copyright Office has stated that protection covers elements of “human authorship” (source: U.S. Copyright Office, 2023–2024 policy). The line between tool and author is shifting: how much human input is enough to turn a generation into a work?

For AI, the impact cuts both ways. Frontier models will build in provenance by design: immutable metadata, cryptographic markers (C2PA), automatic disclosure—to head off more “Schloms cases.” The European Commission’s Code of Practice on Disinformation and guidance on synthetic content point that way; major vendors are adopting C2PA and neural watermarks (sources: European Commission; C2PA). Museums and archives, meanwhile, will use AI for augmented authentication: spectral analysis, stratigraphic reconstructions, assisted style models—extending, not replacing, curatorial expertise.

What’s next? By late 2026, three trends are plausible: standardized labels for synthetic art with penalties for commercial watermark removal; new licensing deals for datasets curated by cultural institutions; and the rise of human–machine co-signatures, with explicit, auditable creative responsibility. At that point, “Who is the artist?” will be less ideological and more evidentiary: process logs, prompt audits, provenance chains.

So the Schloms–Monet episode isn’t a prank so much as a mirror. It reminds us we see what we expect to see. If a label can demote Monet, it can also ennoble mediocrity. Our task is to build tools and rules that make labels truthful, verifiable, contextual—not to vitrify art, but to preserve its promise: to find, in a mark, someone’s presence. Tomorrow, that someone may be plural—the artist, the machine, and the public that chooses what deserves the name of art.