'DSA design a factor in harmful online content'

Harmful content exists online because of article 8 of the Digital Services Act (DSA) and financial gain, a Media, Internet and Data Protection Bar Association (MIDBA) event has heard.  

Harmful content includes bullying/ harassment, suicide/self- harm, hate speech, terrorism, child sexual abuse material (CSAM), and fake accounts. 

Article 8 of the DSA governs “intermediary service providers” and covers any platform that hosts user-generated content. 

Speaking at How to Deal with Harmful Content Online (4 November), barrister Michael O’Doherty explained that article 8 of the DSA meant that platforms did not have to, and could not be obliged to, monitor information before it was uploaded. 

'Political show' on TikTok 

In his opinion, the recent summoning of TikTok before an Oireachtas committee over joyriding videos was a political show. 

“Why drag TikTok in to explain why the videos are online when all you have to do is look at the legislation”, he said. “The videos are on TikTok because TikTok does not have to proactively monitor”. 

However, he added, that did not mean that the platform did not moderate.  

Article 6 provides a hosting defence, where a provider lacks actual knowledge of illegal content or activity and, once notified, acts expeditiously to remove it.

Article 7 encourages voluntary moderation without providers automatically losing the hosting defence. 

AI replacing moderators 

O'Doherty observed that platforms were scaling back proactive moderation – for ideological reasons in the case of X and for cost reasons elsewhere – as human moderators were replaced by AI. 

Protections offered to the platforms have their origins in the E-Commerce Directive 2000 and its aim of keeping the internet operating smoothly without over-litigation. 

O’Doherty said this was important but also pointed out that harmful content could, nevertheless, generate revenue. He cited a November 2025 AML Intelligence report estimating that fake and scam advertisements generated $16 billion a year for Meta. 

'Preserve evidence' 

The first step in getting content removed is to preserve evidence. O’Doherty advised that content might disappear, so practitioners should take screenshots immediately, copy the URL and retain it in a document. The URL is particularly important if litigation follows. 

They should then: 

• Check the platform’s terms of service and community standards to identify the rules allegedly breached, 
• Contact the person responsible for the account and request removal, keeping a record of the request and any response, 
• Report the material through the platform’s online complaints mechanism, and 
• If the complaint is rejected, use the platform’s internal appeal process. 

He stressed that every stage should be documented. Screenshots should be taken of the original complaint, subsequent appeals, and all responses from the platform – including messages that appeared only briefly on screen. 

If a complaint is upheld, the platform may issue a warning, remove the post, suspend the account, or disable it permanently.  

However, the absence of verified identity checks can make repeat offending easy. O’Doherty cited a 2021 case in which a young person harassed a BBC Northern Ireland journalist through X, creating 40 accounts; the platform disabled accounts on 150 occasions. 

Statutory routes 

Several statutory routes may be available, depending on the nature of the material. 

Urgent injunctions may be sought under section 33 of the Defamation Act 2009, section 117 of the Data Protection Act 2018, and sections 40(5A) and 127 of the Copyright and Related Rights Act 2000. Compensation may also be available for defamation, data-protection, or intellectual-property infringements. 

There is no stand-alone tort of harassment simpliciter in Irish law, but claims may be possible through the constitutional right to privacy, applying the principles in Sullivan v Boylan.  

O’Doherty suggested that civil restraining orders, which include communications threatening a victim’s safety or welfare, could potentially be used in online harassment cases. 

He noted that identity theft “doesn't seem to be actionable. It is what they do with that identity that could give you the cause of action” – potentially as a breach of privacy or a breach of data-protection rights.  

Criminal remedies include section 10 of the Non-Fatal Offences Against the Person Act 1997, and sections 2, 3, and 4 of the Harassment, Harmful Communications and Related Offences Act 2020. 

Identification orders 

Where the author is anonymous, a section 45 identification order, or a Norwich Pharmacal Order may be available. 

O’Doherty flagged that the identification order was limited to defamation and the requirements, particularly that it was “likely to succeed at trial”, set a high threshold, whereas Norwich Pharmacal relief is available against any form of unlawful activity online. 

Where proceedings against the author are impractical, the platform itself may be the more realistic target. Under the DSA, however, the hosting defence means that the timing and content of notification are important. 

O’Doherty concluded with what he stressed was his own view on pursuing a platform. Write to its legal department and: 

• List the URL of each item, 
• Explain how the content affects the client’s rights, 
• Describe the use of the internal complaints process, 
• State the date on which the platform acknowledged the complaint, which, in his opinion, is when the “expeditious” clock starts, 
• Describe any attempt to contact the author, or the belief that they are anonymous, 
• Request confirmation within ten working days that the content has been removed, failing which a court application will follow, and 
• Where the harm is serious, the matter urgent, and there has been no delay, indicate that interlocutory relief may be sought.