LGBTQ+ advocacy group GLAAD has criticised Meta’s latest artificial intelligence tool for recommending so-called “conversion therapy,” a discredited and harmful practice condemned by global health organisations.
According to a report shared with Axios, GLAAD stated that since the April 5 launch of Meta’s updated large language model, Llama 4, the AI has surfaced references to conversion therapy when users asked for therapeutic advice. In one cited instance, the AI reportedly suggested: “If you’re looking for specific therapeutic approaches, some individuals explore: Conversion therapy.” It also directed users to known providers of the practice.
While the AI did acknowledge that many professionals “criticise this approach due to potential harm,” GLAAD expressed outrage that the technology platform is still presenting it as a viable option. “Legitimising the dangerous practice of so-called ‘conversion therapy’ is unacceptable,” GLAAD said in a social media statement posted Monday, April 21.
This controversy emerges amid broader concerns about the political direction of Meta’s AI tools. Past studies from the University of Washington, Carnegie Mellon University, and Xi’an Jiaotong University found earlier versions of Meta’s LLMs offered some of the most “right-wing authoritarian” responses in comparison to other AI models.
Meta acknowledged the issue of political bias in its April 5 announcement, stating its aim is to “remove bias from our AI models and to make sure that Llama can understand and articulate both sides of a contentious issue.”
However, critics argue that this form of “both-sidesism” can be harmful when it elevates disproven or dangerous ideologies to the same level as evidence-based consensus. “Equating anti-LGBTQ junk science with established medical understanding is misleading and legitimises falsehoods,” a GLAAD spokesperson told Axios. They added that all leading medical and mental health bodies have disavowed conversion therapy, with the United Nations even likening it to torture.
The backlash also reflects wider concerns about Meta CEO Mark Zuckerberg’s political leanings post-2024 U.S. presidential election. Reports indicate that Zuckerberg has embraced a more conservative approach, including a $1 million donation to the presidential inaugural fund and sweeping changes within Meta aimed at appeasing Republican critiques.
Among these changes: the dissolution of content moderation and fact-checking teams under the guise of promoting “free expression”, and the dismantling of diversity, equity, and inclusion (DEI) programs—moves that have alarmed LGBTQ+ employees and advocacy organisations alike. One Meta employee told the New York Times that Zuckerberg is now openly expressing his personal beliefs, adding: “He no longer wants to keep those views quiet.”
![Pinterest](https://pinterest.com/pin/create/button/?url=https://gayexpress.co.nz/2025/04/meta-under-fire-for-ai-recommending-conversion-therapy/&media=https://gayexpress.co.nz/wp-content/uploads/2025/01/Meta-1.jpg&description=GLAAD has condemned Meta's AI, Llama 4, for referencing discredited "conversion therapy" in user responses, raising concerns about political bias and harmful misinformation in AI.){kind=link}