Netflix and the Reach of Reputation: a Taiwan Trade Mark Decision and the South African Perspective

A recent decision of the Taiwan Intellectual Property Office (“TIPO”) illustrates how trade mark protection operates where competing marks consist essentially of stylised single letter devices, and where the earlier mark has acquired substantial recognition through extensive use. The decision, issued on 27 August 2026 in Opposition no. G01150059, concerned Taiwan registration no. 02496095 for a stylised letter “M”, owned by Mio Media Co., Ltd., which was opposed by Netflix, Inc. TIPO ultimately ordered that the registration be cancelled under Article 30(1)(10) of the Taiwan Trademark Act, having regard to the similarity between the marks, the identity or high degree of similarity between the relevant services, and the considerable recognition and distinctiveness that Netflix had established in its particular stylised “N” device.

Netflix, Inc.’s trade mark	Mio Media Co., Ltd’s trade mark

 

The decision does not concern exclusive rights in an alphabetic character in the abstract. It concerns the protection attaching to a particular stylised device, assessed as a whole, where extensive use has given that device substantial source-identifying significance. It also provides a useful comparison with South African trade mark law, although the comparison needs to be made with some care. Taiwan ultimately dealt with the opposition under a conventional likelihood of confusion provision, whereas South African law contains both conventional deception/confusion-based protection and a separate anti-dilution mechanism for certain well-known registered marks. The respective statutory frameworks therefore raise related, but not identical, questions about the protection of reputation and distinctiveness.

The Taiwanese decision

The contested mark was a stylised capital “M”, presented through a three-dimensional graphic treatment using pink and purple-red gradient blocks. It was registered in class 41 in respect of a substantial range of entertainment and audiovisual services, including online entertainment, online video viewing, non-downloadable films and television programmes, film production and distribution, scriptwriting and adaptation, directing, drama and theatre production and film studio services.

Netflix opposed the registration under Articles 30(1)(10) and 30(1)(11) of Taiwan’s Trademark Act.  Article 30(1)(10) is concerned with the conventional likelihood of confusion analysis, where an earlier mark and a later mark are identical or similar and are used in relation to identical or similar goods or services, such that there exists a likelihood of confusion amongst relevant consumers.  Article 30(1)(11) addresses the protection afforded to a well-known trade mark and applies where the later mark gives rise either to a likelihood of confusion or to a likelihood of dilution of the distinctiveness or reputation of the well-known mark.

The proprietor of the M mark did not file a defence within the prescribed period, and TIPO ultimately determined the matter under Article 30(1)(10), making it unnecessary to decide the alternative ground under Article 30(1)(11).  Unlike the South African trade mark system in which opposition proceedings take place before registration, Taiwan’s trade mark system provides for opposition proceedings against a trade mark after its registration has been published, with the result that a successful opposition leads to the cancellation of the registration.

The cancellation of the M registration was not based simply on the fact that Netflix’s N device was well known.  Rather, the reputation and distinctiveness of the earlier mark formed part of the factual matrix against which TIPO assessed the likelihood of confusion. The earlier rights relied upon by Netflix were a number of registered stylised N devices and the comparison was consequently between the particular graphic representations of Netflix’s N and Mio Media’s M, rather than between the abstract concepts of the letters N and M.  TIPO considered the marks to possess a meaningful degree of similarity, identifying a number of common visual characteristics, including the presentation of a single capital letter, colouring within the red spectrum, a three-dimensional or folded, ribbon-like treatment, relatively flat upper portions and somewhat curved lower portions. TIPO also referred to phonetic similarity between the letters. The assessment therefore proceeded from the overall impression created by the respective devices and the manner in which their individual elements combined, rather than from any proposition that the underlying letters themselves were capable of exclusive use.

The services strengthened the case considerably. The M registration covered a broad range of entertainment and audiovisual services, including online entertainment, online video viewing and film and television-related services, which TIPO regarded as identical or highly similar to services with which Netflix was associated. Film studio services were also considered sufficiently related to Netflix’s activities in video production and audiovisual recording.

TIPO also considered the recognition and distinctiveness of Netflix’s N device. Netflix submitted extensive evidence concerning its global subscriber base, brand recognition, original programming, presence in the Taiwanese market, Taiwanese productions, digital platforms and social media presence, together with other evidence intended to demonstrate the extent to which the Netflix N device had become familiar to the relevant public. TIPO accepted that evidence and regarded the N device as possessing considerable distinctiveness and recognition in Taiwan.

TIPO ultimately concluded that, taking the similarity between the marks, the identity or high degree of similarity between the services, and the reputation and distinctiveness of Netflix’s N device into account, consumers exercising ordinary care may be led to confuse the source of the services or to believe that there was a commercial relationship between the undertakings. The M registration was consequently cancelled under Article 30(1)(10).

The decision is therefore better understood as an illustration of the protection afforded to a distinctive and well-known device.  The factors that proved significant were the overall presentation of the mark, the distinctiveness it had acquired through use and the extent to which consumers had come to associate it with Netflix and its services.  These considerations informed TIPO’s assessment of similarity and likelihood of confusion and ultimately supported the cancellation of the M registration.

The South African position

The South African comparison becomes particularly interesting because the Trade Marks Act 194 of 1993 provides more than one route through which the proprietor of a strong or well-known mark may obtain protection, and those routes should not be conflated.  Section 10(14) provides the conventional form of protection against a later mark which is identical or similar to an earlier registered mark where the relevant goods or services are the same or similar and the use of the later mark would be likely to deceive or cause confusion. The enquiry under this provision is therefore concerned with the familiar question of whether the relevant consumer is likely to be misled as to the origin or commercial connection of the goods or services. Section 10(14) does not make the reputation or well-known status of the earlier registered mark a requirement.  Whether, and to what extent, the reputation of that mark may nevertheless be considered in assessing the likelihood of confusion has not been conclusively settled by our Courts.  In Pepsico Inc v Atlantic Industries, the Court questioned whether the existence and reputation of the earlier PEPSI mark could be taken into account in the comparison under section 10(14), given the principle that the marks must be assessed by reference to their registered form and the fair and normal uses to which they may be put.  The Court left the question open.  The distinction from section 10(17) is consequently clear at the level of the statutory requirements, although the precise role of reputation within the conventional confusion enquiry remains a separate question.

Section 10(17), by contrast, addresses a materially different form of harm, and applies where a later mark is identical or similar to a trade mark that is already registered and well known in South Africa, and where the use of the later mark would be likely to take unfair advantage of, or be detrimental to, the distinctive character or repute of the registered trade mark. The provision expressly applies notwithstanding the absence of deception or confusion.

There is a procedural qualification to this comparison. Section 10(17) is commonly encountered as a ground of opposition to registration, and the Supreme Court of Appeal has described it as a self-standing anti-dilution ground at the registration stage. Section 10 itself, however, expressly provides that the marks to which its prohibitions apply may, if already registered, be liable to removal from the register. Accordingly, a registration alleged to offend section 10(17) may be challenged by way of an application for removal under section 24 of the Act.  The position is distinct from section 34(1)(c), which provides the corresponding anti-dilution protection in the context of infringement.  The Supreme Court of Appeal has recognised section 10(17) as an anti-dilution provision, and the corresponding infringement provision in section 34(1)(c) has similarly been interpreted as protecting the reputation, advertising value and selling power associated with a well-known mark independently of the conventional likelihood-of-confusion enquiry. In Yuppiechef Holdings (Pty) Ltd v Yuppie Gadgets Holdings (Pty) Ltd, the Supreme Court of Appeal made clear that similarity under section 34(1)(c) is to be considered without reference to the potential for confusion or deception, because the statutory harm addressed by the provision is of a different character.

The distinction was again relevant in National Brands Limited v Cape Cookies CC and Another, in which the Supreme Court of Appeal identified section 10(17) as an anti-dilution provision and distinguished it from the more conventional provisions concerned with deception and confusion. The Court also emphasised that section 10(17) is a self-standing ground of opposition, rather than merely a fallback where a conventional confusion-based ground cannot be established.

The South African authorities accordingly recognise section 10(17) as a true anti-dilution provision. Unlike conventional confusion-based provisions, the enquiry is not whether consumers will mistake the later mark for the earlier one, but whether the later use is likely to take unfair advantage of, or be detrimental to, the distinctive character or repute of the registered mark.  That distinction should not, however, be expressed as a general proposition that confusion is irrelevant whenever a mark is well known. The South African legislation contains several different provisions dealing with well-known marks, and their requirements differ. The absence of a confusion requirement is particularly characteristic of the anti-dilution route under section 10(17), rather than a universal feature of every well known-mark claim.

The distinction matters for the Netflix comparison. Under the conventional South African registration ground in section 10(14), the enquiry would resemble the Taiwanese Article 30(1)(10) analysis: the court would consider the similarity between the marks, the relationship between the relevant goods or services and the likelihood of deception or confusion. The strength and reputation of the earlier mark would form part of that overall assessment. Under section 10(17), however, the enquiry would be different. Confusion or deception is expressly unnecessary; the question would instead be whether the later mark is identical or similar to the well-known registered mark and whether its use would be likely to take unfair advantage of, or be detrimental to, the distinctive character or repute of that mark. The degree of similarity between the marks is merely a factor in assessing the probable impact that the use of the impugned mark will have on the well-known mark.

The relevant enquiry under section 10(17) would not simply be whether Netflix is a famous corporate entity or whether the NETFLIX brand is internationally recognised. Rather, the question would be whether the N device relied upon by Netflix is a registered trade mark and is well known in South Africa. One should therefore distinguish between the reputation of Netflix as a corporate entity, the reputation of the NETFLIX brand and the recognition of the particular N device. These considerations inform the application of section 10(17) which requires the assessment to focus on the registered mark and the potential for the later mark to take unfair advantage of, or be detrimental to, its distinctive character or repute.

If the N device qualifies as a registered trade mark that is well known in South Africa, the focus shifts to the relationship between the respective devices. The next enquiry under section 10(17) would be whether the later M device is identical or similar to it and whether use of the M would be likely to take unfair advantage of, or be detrimental to, the distinctive character or repute of Netflix’s registered trade mark. Because confusion or deception forms no part of that statutory test, the enquiry would differ materially from that undertaken by TIPO. The visual similarities identified in the Taiwanese decision could remain relevant to the question of similarity, but the South African analysis would then ask what the use of the M does to the distinctive character or reputation of the N, or whether it appropriates some of the commercial value associated with the N, rather than simply asking whether consumers are likely to believe that the M originates from Netflix or from an undertaking commercially connected to Netflix.

The most useful comparative conclusion is not that Taiwan and South Africa would necessarily reach the same result, but that the same factual dispute may engage different legal interests depending upon the statutory route relied upon. In Taiwan, the M device was cancelled because TIPO considered it sufficiently similar to Netflix’s N devices in relation to identical or highly similar services and in circumstances where the recognition of the N contributed to a likelihood of confusion. In South Africa, the same facts could potentially support either a conventional confusion-based claim or, where the statutory requirements are satisfied, an anti-dilution claim under section 10(17).

The strength of a trade mark lies not merely in the word, letter or symbol from which it is constructed, but in the distinctive commercial identity that consumers associate with the sign as a whole. As that identity becomes more established, the law seeks to balance two competing interests: the freedom of other traders to develop and use their own brands, and the legitimate interest of the trade mark proprietor in preserving the distinctive character and reputation embodied in its mark. The divergent approaches reflected in the Taiwanese and South African regimes demonstrate how different legal systems may protect those interests through different statutory mechanisms.

*The TIPO decision is available here.