Summary
- GLAAD reports that AI technologies are inflicting significant harm on LGBTQ individuals through bias, misinformation, and discrimination.
- The organization cautions that autonomous AI could perpetuate these issues in areas such as housing, employment, healthcare, and lending.
- The report emphasizes the need for enhanced oversight, better training datasets, and collaboration between AI firms and civil society organizations.
According to a recent report from GLAAD, artificial intelligence is exacerbating anti-LGBTQ biases, misinformation, and discrimination, impacting various sectors like healthcare, employment, housing, and privacy.
The report, titled “Build for Everyone: A Framework for LGBTQ Representation and Safety in AI,” was published on Wednesday and asserts that ensuring LGBTQ safety should be a fundamental principle in the ethical development of AI.
GLAAD warns that AI systems that rely on biased or incomplete datasets risk reinforcing stereotypes, marginalizing LGBTQ voices, compromising user privacy, and leading to discriminatory results as these technologies become more integrated into daily life.
“AI is a civil rights issue,” stated GLAAD President and CEO Sarah Kate Ellis in the report. “Neutrality is no longer an option. To build AI that is ethical, inclusive, and responsible, tech leaders must proactively embrace intentional practices to create safe products.”
The report identifies several areas of concern, including biased training data, misinformation targeting LGBTQ individuals, discriminatory results from predictive AI systems, failures in content moderation, and privacy vulnerabilities, arguing that incomplete or inaccurate information about LGBTQ communities can perpetuate harmful stereotypes.
Ellis emphasized that responsible AI development is not only beneficial for businesses but also essential for the sustainability of AI companies. “More than 20 percent of Gen Z identifies as LGBTQ,” she noted. “These individuals represent your future employees and customers.”
A 2023 study from LGBT Capital found that LGBTQ individuals hold a global purchasing power of $4.7 trillion, a figure projected to rise to $33 trillion by 2030.
“To contextualize this, if we were a separate nation, we would rank as the fourth-largest economy globally,” Ellis remarked.
The report emerges amid ongoing discussions regarding AI bias. In May, researchers discovered that prominent AI models showed a consistent bias favoring Catholicism, while responding less favorably to Jehovah's Witnesses, atheism, and agnosticism.
Earlier this month, Devin Kim, a former engineer at xAI, filed a lawsuit against xAI and SpaceX, claiming he was terminated for raising concerns about Grok's insufficient safeguards against misinformation and bias. Concurrently, xAI, led by Elon Musk, is engaged in a legal dispute with Colorado over a state law mandating companies to evaluate and mitigate discrimination risks in AI systems used for decisions in housing, employment, and lending.
GLAAD contends that the repercussions of AI bias extend beyond interactions with chatbots and image generation tools.
“While not exclusively affecting LGBTQ people, we must highlight other emerging challenges as AI development progresses,” the report indicated. “These issues may include model hallucinations or sycophantic behavior that generates misinformation on critical topics like health and elections.”
As companies advance toward AI agents capable of executing tasks with minimal human intervention, GLAAD cautioned that such autonomous systems could inherit existing biases and automate discriminatory practices, such as excluding LGBTQ-friendly healthcare providers from search results or making erroneous assumptions about users’ identities.
To mitigate these risks, GLAAD urges developers to enhance LGBTQ representation in training datasets, bolster privacy protections, maintain human oversight in moderation systems, and collaborate more closely with advocacy organizations. The report also calls for greater accountability within the industry and stronger regulatory measures.
“Neglecting to incorporate LGBTQ experiences and issues in training datasets, product design, and governance can lead to harm for marginalized communities and result in inferior, less trustworthy products that may alienate an expanding demographic of users,” the report concluded.