TORONTO, June 30, 2026 (Globe Newswire) — Vector Institute launches today unbiased plusis a free, open-source AI tool built by Vector’s AI safety research scientists that helps you detect, explain, and rewrite biased language in written content and AI training datasets in seconds.
Release comes at a decisive moment. canadian National AI Strategy: AI for everyone Recognizing that unchecked bias in AI systems undermines fairness, safety, and public trust, it makes clear that addressing algorithmic bias is a pillar of responsible AI development. The scale of the challenge is staggering. Bias in AI training data is measured as a percentage in the following ranges: 3.4% to 38.5%even safely calibrated large-scale language models (LLMs) exhibit implicit racial and gender bias. This is a worrying reality, given that LLMs risk perpetuating and amplifying the very human biases embedded in the texts they study.
UnBias-Plus is the first free tool of its kind available to both organizations and the public. The tool analyzes text written for biased language across different dimensions of race, gender, age, and political frameworks, explains exactly why the language was flagged, suggests neutral alternatives, and returns a completely rewritten version of the original content. Rather than simply creating a revision, UnBias-Plus shows users why language is biased and helps them more clearly understand how bias manifests in everyday communication. The platform is available in two formats.
●A Browser-based version Intended for public use and supports content up to 750 words. It is practical for everyday writing and for organizations evaluating tools before widespread adoption. and
●A developer installer Integrate into enterprise applications, systems, and workflows.
“Our inspiration for building this was simple: the people most harmed by prejudiced language are often the last to know it exists. Patients are unaware of the assumptions buried in their clinical records. Candidates wonder why doors are closed. We wanted to make the invisible visible: a free tool to help surface, explain, and correct bias. But bias doesn’t just exist in someone’s writing today. It exists in the data used to train AI models, where it is learned, replicated, and deployed at scale across the organization.UnBias-Plus enables developers and data scientists to find and remove bias before it becomes part of the model.”
—Shaina Raza, PhD Applied Machine Learning Scientist, Vector Institute Responsible for AI
UnBias-Plus is built to help professionals and fields where language bias has the most significant real-world consequences.
editorial and media – Editors can integrate UnBias-Plus directly into their workflows to identify and correct bias before publication, supporting more accurate and comprehensive reporting.
HR, EDI, Workplace Team – Use this tool to help filter out gender, racial, and other language in job descriptions, performance reviews, promotion stories, and internal communications that may unintentionally create barriers for employees and candidates.
healthcare communication – UnBias-Plus helps you review clinical records, care assessments, and patient communications for bias and discriminatory frameworks that can impact care decisions, insurance coverage, and outcomes. For example, if a caregiver writes, “A noncompliant patient refused medication,” that statement will stick with the patient at every subsequent visit. UnBias-Plus flags such language and may suggest alternatives, such as “the patient refused medication,” so the record remains factual rather than judgmental.
insurance – The team helps screen benefit letters, eligibility letters, and media coverage to ensure language reflects policy and facts, rather than unconscious assumptions about who should receive care.
AI developers and data scientists – UnBias-Plus helps with annotations, prompts, and screening of source text. This allows sanitized data to be used to train or fine-tune AI models, identifying biased patterns in the data layer and then extending those patterns to trained, replicated, and deployed systems. AI teams can also run model output through tools and audit it for biased language before it reaches users or is embedded in downstream workflows.
“Canada’s AI leadership depends on building systems that produce more equitable outcomes, and that means making sure the research behind those systems is accessible to everyone. Making UnBias-Plus free, transparent, and open source is an intentional expression of Vector’s mission as an independent, not-for-profit institution. AI “By sharing this research with developers, organizations, and the public, we are ensuring the ability to build more equitable AI belonging.”
— Kathryn Hume, Vector Institute AI Engineering Vice President.
The browser-based version of UnBias-Plus is currently available at: unbias-plus.vectorinstitute.ai/. The open source developer installer is available at: pypi.org/project/unbias-plus/.
UnBias-Plus is the first release of Vector Institute’s broader suite of Safe AI tools. Updates extend the capabilities of future audio and video bias detection and misinformation detection tools. Visit Vector’s site to access other available products.
UnBias-Plus currently only supports English text. This tool does not verify factual accuracy, detect misinformation, or identify AI-generated content.
About Vector Institute
Vector Institute for Artificial Intelligence is an independent, non-profit organization dedicated to AI research with an emphasis on machine learning and deep learning. Based in Toronto, Ontario, Vector attracts world-class researchers and works with industry, academia and government to advance AI innovation and talent development in Canada. For more information, please visit vectorinstitute.ai.
