An Irish parliamentary committee has released a report recommending new AI legislation regulating “hate” and “misinformation”.
The First Interim Report from the Joint Committee on Artificial Intelligence, published Tuesday, outlines a strict regulatory framework that rejects calls for a lighter touch on tech innovation.
MODERATION OF “HATE” AND “MISINFORMATION”, CITING GOVERNMENT-FUNDED NGOs
Regarding content moderation, the committee states that “harmful and hateful content pushed by recommender systems that use AI needs to be better addressed” in legislation.
“Another regulation gap discussed was harmful and hateful content pushed by recommender systems that use AI,” the report reads.
“This is of serious concern to the Committee and must be addressed in any EU and national legislation.”
Specifically, the Committee cites the Irish Traveller Movement and BeLonG To – two heavily Government-funded Leftwing NGOs that campaign on “anti-racism” and LGBT issues respectively.
“The Committee heard from the Irish Traveller Movement about how Traveller children are particularly vulnerable to hate based harms online, with automated discrimination inherent in some LLMs, algorithms and pages or bots being set up solely to negatively stereotype Travellers,” the Committee states.
“Similarly, BeLonG To told the Committee that LGBT+ young people suffer from AI perpetuating discriminatory stereotypes and from AI-generated content that targets minority groups online. They are troubled by the recent weakening of content moderation by online platforms.
“Supports for underprivileged communities and marginalised groups have a big role to play in addressing these issues. The session with young people underlined the need for strong enforcement of the Digital Services Act and safety by design and highlighted the damage that harmful and hateful content pushed by recommender systems that use AI causes.”
Notably, the Irish Traveller Movement’s ‘Yellow Flag’ programme, which teaches Irish school children about the idea of ‘white privilege’ in classrooms across the Republic of Ireland, just received renewed funding to the tune of €100,000 from the Department of Justice.
The report also calls for specific obligations on platform owners to “prevent the use of AI-driven recommender systems for misinformation campaigns aimed at destabilising society”.
EUROPEAN AI LAWS SHOULD NOT BE “DILUTED”
It advises the Irish government to treat the European Union’s AI Act as a “minimum baseline” rather than a maximum standard , explicitly warning against any attempts to “dilute” the legislation.
“Ireland must not shy away from the EU AI Act or try to dilute it,” it reads.
“We should treat it as a minimum baseline for national AI regulation, not a maximum standard.”
RECOMMENDER SYSTEMS SHOULD PUSH A “BALANCED POINT OF VIEW”
The report focuses heavily on the mechanics of how content is delivered to users. Recommendation 63 states that “recommender systems should be designed so that recommended material that is put out delivers a balanced point of view, that is evidence based”.
The committee further recommends that social media “recommender systems should be switched off by default” for general users. For accounts used by children, the report goes further, stating that social media companies “should be banned” from activating these algorithms entirely.
INNOVATION VS. REGULATION
The report challenges the argument that strict regulation hampers technological progress. In a section titled “Innovation and regulation can coexist,” the committee references testimony from the Irish Council for Civil Liberties (ICCL), which argued that “the false dichotomy of innovation vs. regulation serves the vested interests of billionaires”.
The committee concludes that “robust, well-implemented regulation of AI is essential” and not mutually exclusive with innovation.
AI SHOULD AVOID “OVERRELIANCE ON THE PRIVATE SECTOR”
The report also addresses the relationship between the state and private tech firms. It recommends that the state “take action to mitigate against an overreliance on the private sector” regarding AI and explore “publicly owned AI resources and technologies”.
On the issue of intellectual property, the committee recommends that the EU Copyright Directive be strengthened “to ensure that content cannot be used to train AI models without the consent of its creators”.
The findings will now be considered by the government, with the committee calling for a “national AI Office” to be operational by August 2026 to lead the coordinated response.