Breaking

GLAAD Report Warns AI Is Amplifying Anti-LGBTQ Bias

June 17, 2026 at 09:13 EDT

Policy & Regulation
Foundation Models
Security

LGBTQ advocacy group GLAAD released a new report in June 2026, "Build for Everyone," warning that generative AI has begun to reproduce and amplify anti-LGBTQ bias and misinformation, potentially threatening users' health, safety and civil rights. The report was previewed at the Axios AI+NY Summit, with details reported on June 17.

June 2026 · GLAAD

When AI Learns Bias, LGBTQ+ People Pay the Price

GLAAD's new framework, "Build for Everyone," warns that AI trained on skewed data amplifies misinformation and stereotypes about LGBTQ+ people — with real-world risks to health, safety and civil rights.

60+

countries where being LGBTQ+ is illegal — AI queries could reveal identity and trigger danger

200+

civil society groups signed the April 2024 open letter to Big Tech the framework builds on

major foundation models whose bias ripples across the whole app ecosystem

Why fix the foundation first

A small number of foundation models sit beneath countless apps — correcting bias early has the broadest effect.

4 foundation
models

→

countless downstream apps

Harms the report documents

Meta's Llama 4 recommending conversion therapy — an unscientific, discredited practice
Content moderation wrongly suppressing legitimate LGBTQ+ content
Privacy leaks: a user's questions can reveal sexual orientation or gender identity
Misgendering and stereotyped outputs in everyday generative AI use

The framework's five fixes

Correct bias in foundation models

Avoid automated discrimination

Ensure human oversight

Respect data privacy

Partner with civil society

CEO Sarah Kate Ellis: the group "cannot stay neutral," calling for urgent fixes to biased outputs on healthcare and employment queries. Detailed benchmarks and timelines remain unpublished at the preview stage.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.