The AI graveyard: a running list of projects and startups that didn’t make it

Many AI products that once promised innovation are now failing due to competition, unsustainable costs, or privacy concerns.

Multiple AI projects and startups have recently shut down or faced significant challenges, contributing to a growing "AI graveyard." Relay, an AI workflow automation tool, ceased operations in August due to competition from larger platforms. Character.ai, a chatbot platform, filed for bankruptcy shortly after its five-year launch anniversary. Even established companies are experiencing setbacks; OpenAI discontinued its ChatGPT Atlas web browser and Operator AI agent, folding their features into ChatGPT, and its Sora video platform faced high costs and retention issues before shutting down in March 2026. Apple delayed its enhanced Siri due to engineering issues, leading to a $250 million settlement over AI marketing claims. Microsoft's "Recall" feature faced significant privacy backlash, delaying its release and still raising security concerns. Notion Mail is scheduled to close on September 22, as users opted for separate AI agents. In hardware, Humane AI Pin ceased operations due to performance issues and a fire risk associated with its charging case, with its assets acquired by HP. The Rabbit R1, despite strong initial sales, received poor reviews for performance and limited integrations. Huxe, an AI audio app, shut down in May 2026 due to similar features being offered by larger platforms like Spotify. Yupp, a platform for comparing AI models, closed in March 2026 due to unsustainable costs and a lack of product-market fit. Figgs AI, which allowed users to create AI characters, also shut down because its free service became too expensive to maintain.

The high failure rate in the AI sector indicates significant market saturation and challenges in differentiating unique value propositions, particularly against integrated offerings from tech giants. Startups and even larger companies struggle to find viable business models when core functionalities can be absorbed or replicated by platforms with vast user bases and development resources. This trend suggests a consolidation phase may be approaching, where only the most robust or uniquely positioned AI solutions will survive.

The persistent privacy and security concerns surrounding AI features, exemplified by Microsoft's Recall, highlight a critical hurdle for widespread adoption. Consumers and regulators are increasingly scrutinizing how AI systems handle sensitive data, forcing companies to prioritize security and transparency. Future AI development will likely require a stronger emphasis on privacy-preserving technologies and ethical considerations to build trust and avoid costly regulatory interventions or public backlash.