Part of the AI Regulation News hub.
The Italian data protection authority banned further processing of a news anchor's image in AI-dubbed satirical deepfakes and formally warned the broadcaster for breaching the fairness and data protection by design principles
The broadcaster's defence was that satire is exempt from truth. The Garante accepted the satire framing and still found a breach, because the problem was not the content but how plausible it looked.
Bottom line: Binding measure, in force. A prohibition on further processing under Article 58(2)(f) GDPR plus a formal warning under Article 58(2)(b). No fine was imposed.
Who this affects: Italian media lawyers, broadcasters and production companies using synthetic voice or image, data protection officers at media groups, and anyone advising on satire, likeness and AI disclosure in the EU.
Issue date: 23 July 2026, register of measures no. 577. RTI has 30 days from receipt to report what it has done. Opposition to the ordinary courts within 30 days of communication, or 60 days if the applicant resides abroad.
What changed: A regulator that plainly accepts satire as a protected form of expression under Articles 136 to 139 of the Italian Privacy Code still found the processing unlawful, on the ground that the alteration was not perceptible enough.
Analysis: The finding rests on Article 25 GDPR as much as Article 5. RTI's disclaimers existed. The Garante's objection was that they were calibrated for an attentive viewer with some technological literacy, and that data protection by design required communications explicit enough to reach the inattentive viewer and the one who tuned in mid-programme.
Primary sources: Provvedimento del 23 luglio 2026, doc. web 10281021 · Garante press release, 7 August 2026
- Instrument (EN)
- Measure of 23 July 2026, register no. 577, doc. web 10281021
- Authority
- Garante per la protezione dei dati personali
- Jurisdiction
- Italy
- Status
- Adopted and communicated; not final if opposed within the statutory period
- Bindingness
- Binding on R.T.I. Reti Televisive Italiane S.p.A. Breach of a processing ban can attract the criminal penalty in Article 170 of the Code and the administrative fine in Article 83(5)(e) GDPR.
- Issue date / next deadline
- 23 July 2026. Report on implementation due within 30 days of receipt. Opposition within 30 days, or 60 days for a claimant resident abroad.
- Provisions found breached
- Article 5 GDPR (lawfulness, fairness, transparency) and Article 25 GDPR (data protection by design and by default)
- Measures imposed
- Prohibition on further processing, formal warning, annotation in the internal register under Article 57(1)(u) GDPR
- Primary source
- https://www.garanteprivacy.it/garante/doc.jsp?ID=10281021
The complaint
Enrico Mentana, director of the La7 evening news, complained under Article 77 GDPR that R.T.I. Reti Televisive Italiane S.p.A., as publisher of Striscia la Notizia, had reproduced his image using artificial intelligence to build segments in which he appears, unaware, in his own professional role as a news director, delivering a script written by the programme's authors and carrying the La7 news logo.
The complaint recorded that full clips were later uploaded to the programme's website and social channels and to the Mediaset Infinity platform, and asked for deletion of all the images used.
Mentana argued that the disclaimer on the site works only for an attentive and up-to-date viewer, that many of the social fragments carried no indication of AI use at all, and that user comments calling him stupido, ignorante, ubriaco, pagliaccio, rosicone and vigliacco showed viewers believed the statements were authentic.
What the broadcaster argued
RTI said the audiovisual excerpts were licensed from LA7 under a contract signed on 9 March 2021 and subsequently extended, that LA7 knew the clips would air on Striscia la Notizia, and that regular usage reports drew no objection. It stressed the material was bought and invoiced, not scraped from the web.
Its substantive defence was that the only thing it did was dub Mentana with a voice other than his own using computer systems, which is what an impressionist has always done, only without a human voice actor. Satire, it argued, is exempt from the requirement of factual truth because it works through the implausible and the hyperbolic, citing Cassazione civ., Sez. III, ordinanza of 14 March 2024 no. 6960.
On disclosure, RTI said it had gone beyond what the law required, and noted that under Regulation (EU) 2024/1689 the specific transparency obligation would only become applicable from 2 August 2026, with reduced obligations where content forms part of an evidently artistic, creative, satirical or fictional work.
It also pointed out that the La7 news and the satirical programme aired at almost the same time, so Mentana could not have been live on two channels at once.
Why the Garante still found a breach
The authority agreed on the legal frame. It held the processing falls within the exercise of freedom of expression, so Articles 136 to 139 of the Privacy Code apply, and it treated the right of satire as one of the forms of expression those provisions cover.
The finding then turned on perception. The Garante observed that the complainant's real image had been used, filmed inside the television studio from which he directs an evening news programme, and that its genuineness was altered by a dubbing carried out with tools that did not make the alteration clearly perceptible. The statements attributed to him appeared, on immediate viewing, truthful, and were not marked by the degree of exaggeration that would make the falsity obvious.
The real context aggravated it. Putting a working journalist in his own newsroom made the processing particularly insidious for his personal identity and amplified the risk of turning him into a vehicle of disinformation, a risk the Garante found confirmed by the volume of mocking and offensive messages he received.
That, in the authority's view, was not consistent with the fairness principle, which requires a controller to weigh at the design stage the risk of prejudice that a processing operation or the modalities chosen for it may cause to the data subject.
The measures, and what was not imposed
Under Article 58(2)(f) GDPR the Garante prohibited further processing of the complainant's data in the manner described, excepting mere retention for possible use in judicial proceedings. Under Article 58(2)(b) it issued an ammonimento for failure to observe the rules on processing, with particular regard to the measures needed to secure lawfulness, fairness and transparency. The measures were annotated in the internal register under Article 57(1)(u).
No administrative fine was imposed. The authority reasoned that the question was legally novel, arising from recently introduced technological tools whose contours are unsettled, especially where they are used inside entertainment programmes and so belong to artistic expression, and that a warning was the proportionate response.
RTI is invited under Article 157 of the Code and Article 58(1)(a) GDPR to report within 30 days of receipt what it has done to implement the measure. Failure to reply attracts the fine under Article 166 of the Code and Article 83(5)(e) GDPR. The measure is open to opposition before the ordinary courts.
The transferable test
If you advise a broadcaster, note what did not save RTI. A licence to the underlying footage did not. The complainant argued that a licensor can transfer clips but cannot authorise their modification to the detriment of the person filmed; the Garante did not rule on the licence at all, and decided the case on the perceptibility of the alteration and on data protection by design. Prior press coverage did not. A disclaimer in the title, notes under each video and a crawling caption on the platform did not.
What the Garante asked for was disclosure calibrated to the least attentive member of the audience, made evident at multiple points during transmission so that a viewer joining mid-programme still encounters it. That is a design requirement, not a labelling requirement, and it is the part that travels beyond Italy.
What we did not verify
We opened the full Italian text of provvedimento n. 577 of 23 July 2026 at doc. web 10281021 and took every fact and quotation from it. Personal details and several dates are redacted as XX in the published text.
We did not open the Garante press release of 7 August 2026 that the measure cross-references, we did not see the RTI defensive memoranda or the 27 February 2026 hearing record other than as summarised in the measure, and we did not review the videos at issue.
We refuse to say the Garante ruled on defamation, on the right to image under Article 10 of the Civil Code, or on Article 97 of the copyright law. RTI argued those fall outside the authority's competence, and the measure decides the data protection questions only. We also make no claim about whether RTI has appealed.
A disclosure defence is judged against the viewer least equipped to read it, not the viewer best equipped. If your client uses synthetic voice or likeness in an entertainment context, the record you want is not a single disclaimer but evidence that the labelling was designed for a distracted, mid-programme audience and repeated across every surface where the clip lands. Licensing the source footage does not license the alteration.
Source File
https://www.garanteprivacy.it/garante/doc.jsp?ID=10281021
Open doc. web 10281021 on garanteprivacy.it and confirm the register number 577 and date of 23 July 2026, the operative paragraphs a), b) and c) at the foot of the measure, the finding that the disclaimers were insufficient for an average or inattentive audience, and the reason given for choosing a warning over a fine.
il trattamento effettuato con le caratteristiche descritte, idonee, di per se, a far percepire veritieri i servizi trasmessi, non risulta accompagnato dall'uso di disclaimer sufficientemente chiari, quanto meno con riferimento alle conoscenze tecnologiche di un pubblico medio o comunque poco attento ยท Garante per la protezione dei dati personali, provvedimento n. 577, 23 July 2026
FAQ
Was RTI fined?
No. The Garante imposed a prohibition on further processing and a formal warning, and explained that the novelty of the legal question made a warning the proportionate measure.
Did the Garante reject the satire defence?
Not as a matter of principle. It accepted that satire falls within the freedom of expression regime in Articles 136 to 139 of the Privacy Code. It found the specific processing unlawful because the alteration was not clearly perceptible.
Does the licence from LA7 matter?
The complainant argued a licensor of footage cannot authorise its modification to the detriment of the person filmed, and the measure does not treat the licence as curing the data protection breach.
What does RTI have to do now?
Stop the processing in the manner described, except for retention for possible judicial use, and report to the Garante within 30 days of receipt on the steps taken.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.