As artificial intelligence becomes woven into everyday life, from search engines to job applications and digital assistants, LGBTQ people could be paying a disproportionate price for systems that weren’t built with them in mind.
On June 17, GLAAD released its inaugural AI safety report, Build for Everyone: A Framework for LGBTQ Representation and Safety in AI, calling on tech companies to rethink how artificial intelligence is designed, trained, and deployed. The report argues that AI is not neutral by default, and without intervention, existing online bias risks becoming automated at scale.
The report arrives at a moment when AI tools are increasingly shaping how people discover information, communicate, access services, and participate online. GLAAD says that influence comes with responsibility.
Drawing from academic research, journalism, civil society reporting, and GLAAD’s own monitoring efforts, the organization examined the entire AI pipeline, from foundation model training and privacy practices to moderation systems and emerging agent-based technology.
The takeaway is direct: if bias enters the system early, it doesn’t stay there.
Why GLAAD Is Raising The Alarm
At the center of the report is a warning that AI models can absorb and reproduce harmful narratives about LGBTQ communities.
Among the examples highlighted is GLAAD’s previous research into Meta’s Llama models, which found responses recommending conversion “therapy” when users searched for language around “unwanted same-sex attraction,” terminology commonly used by promoters of the practice.
Conversion “therapy” has been condemned by major U.S. medical and mental health organizations and has been compared to torture by the United Nations.
GLAAD President and CEO Sarah Kate Ellis said the industry can no longer treat these outcomes as accidental.
“Neutrality is no longer an option. AI systems trained on data that wrongfully positions LGBTQ lives as ‘fringe’ or treats our equal rights as ‘controversial,’ or that fail to catch sophisticated disinformation about LGBTQ people, threaten our health, safety, and civil rights.”
Ellis added that building safer systems is both an ethical obligation and a long-term business decision.
The organization has increasingly focused on technology accountability in recent years through initiatives including its annual Social Media Safety Index, which evaluates how platforms protect LGBTQ users online.
Four Areas GLAAD Wants Tech To Fix
The report outlines several recommendations aimed at AI developers, deployers, and designers.
Training Data Must Keep Up
GLAAD argues AI companies should regularly update datasets and retrain systems to reflect changes in language, culture, misinformation tactics, and evolving forms of hate.
The report points to prior research showing that language models can reproduce harmful stereotypes and discriminatory associations if left unchecked.
Privacy Can’t Be Optional
Another major concern centers on data collection.
GLAAD warns LGBTQ users may face elevated risks when AI systems collect or infer details related to sexual orientation or gender identity. Those risks become more severe in countries that criminalize same-sex relationships and in places where legal protections are limited.
The report also notes that people increasingly share deeply personal questions with chatbots, including conversations about identity and coming out, raising questions around storage, retention, and user control.
Content Moderation Needs More Nuance
GLAAD says automated moderation tools often miss harassment while also restricting legitimate LGBTQ content.
The report cites ongoing concerns from creators and users across major platforms who say automated systems have contributed to shadowbanning, demonetization, content removals, and account suspensions.
The concern isn’t whether moderation should exist, it’s whether platforms can build systems that recognize context instead of flattening it.
Oversight Matters
The final recommendation pushes for stronger industry accountability and collaboration with communities most affected by AI systems.
Rather than treating safety as a patch after launch, GLAAD argues LGBTQ voices should be included earlier in development and testing.
Building The Next Version Of The Internet
GLAAD Vice President of Research Brandon Grabowski framed the moment as one that is still open for influence.
“While the conversation can make the AI-powered future seem inevitable, there is still time to direct the technology: how safe it is, how accurate its answers are, how appropriate its actions are, and whether the technology is a net-positive or a net-harm to society.”
He added:
“With a few thoughtful protections, we can build an AI future that’s better and safer for everyone.”
That idea sits at the center of Build for Everyone: AI is no longer emerging technology. It’s infrastructure.
And if these systems are helping shape culture, access, opportunity, and information, GLAAD’s message is clear, everyone should be considered before the code ships.
