Inspiration The rapid proliferation of AI-generated content (with over 20% of online material now estimated to contain AI-produced elements) made it clear that people and organisations needed reliable tools to tell human from machine. I saw real harm emerging: misinformation spreading through deepfake video, voice cloning enabling fraud, and AI-written text sneaking into legal documents and job applications. AI Aware was born from the conviction that as AI generation tools become more powerful and accessible, detection capability must keep pace to protect trust, integrity, and public safety.
What it does AI Aware is a unified AI content detection platform that covers four modalities: text, video, images, and audio - all from a single interface. It can identify content generated by major AI tools including ChatGPT, Claude, Gemini, Sora, ElevenLabs, Midjourney, and many others, returning calibrated confidence scores rather than simple binary verdicts. The platform also provides paragraph-level visualisation so users can see exactly where in a piece of content the AI signals appear, enabling informed, contextual decisions rather than blind trust in a percentage.
How I built it AI Aware was built on a proprietary ensemble detection architecture that combines multiple models - each with different strengths, training data, and analytical approaches - rather than relying on a single algorithm.
Challenges I ran into One of the hardest technical challenges was building resilience to "out-of-distribution" content - detecting AI from generators the models had never seen before, as new tools emerge constantly. A growing industry of "AI humaniser" tools specifically designed to defeat detectors meant the team had to go beyond surface-level signals like perplexity scores and train on hybrid and manipulated content. Balancing high accuracy with a very low false positive rate (below 1 in 1,000 for text at document level) was a continuous challenge.
Accomplishments that I'm proud of AI Aware has become one of the only platforms in the world offering multimodal AI detection - text, video, audio, and images - within a single unified product. The team secured backing from Innovate UK. The platform has now been used to test millions of pieces of content, establishing AI Aware as a credible and recognised player in the AI detection space.
What I learned Building AI detection revealed how quickly the threat landscape shifts. Engaging with industries like legal, education, and financial services also taught the team how differently AI risk manifests across sectors, shaping a more nuanced and use-case-aware product roadmap.
What's next for AI Aware AI Aware plans to expand its API and enterprise integration capabilities, making it easier for high-volume organisations - fraud teams, media companies, large recruiters - to embed detection directly into their existing workflows. T
Log in or sign up for Devpost to join the conversation.