The AI Graveyard: Notable Failures and Lessons from the Industry
3–5 minutes
Relay, an AI workflow tool positioned as a Zapier alternative, ceased operations in August after five years, unable to compete as larger platforms embedded similar automation. Character.ai, a chatbot startup founded five years earlier, filed for bankruptcy. According to S&P Global Market Intelligence, roughly 42% of corporate AI projects are eventually abandoned, often due to insufficient funding, technical hurdles, competitive pressure, scaling difficulties, or weak demand. The resulting landscape of defunct AI initiatives offers a catalog of lessons for the broader industry.
OpenAI’s missteps
OpenAI, despite its valuation as a leading private AI firm, has experienced several setbacks. In July 2024 the company attempted to rebrand ChatGPT as a “super app” by merging the traditional interface with separate Chat, Codex, and Work modes and labeling the original as ChatGPT Classic. User backlash over a confusing layout forced OpenAI to revert the change shortly after. A similar consolidation approach succeeded when the standalone browser ChatGPT Atlas was discontinued on August 9, with its capabilities folded into ChatGPT. The same pattern applied to the Operator agent and DALL‑E, whose image‑generation role was increasingly embedded in ChatGPT. The video platform Sora, launched with high expectations, was shut down in March 2026 after struggling with operating costs and retention issues.
Apple’s Siri AI
Apple announced Apple Intelligence in 2024, promising a revamped Siri that could understand context and coordinate across apps. Repeated delays, attributed to engineering problems and bugs, postponed the feature’s release and eventually led to a $250 million settlement over misleading marketing claims about the iPhone 16’s AI capabilities. The updated Siri finally appeared in the iOS 27 beta in July and began rolling out to English‑language users this month, with plans to add other languages later.
Microsoft’s Recall
Microsoft unveiled Recall at its 2024 Build conference, describing it as an AI‑driven “photographic memory” that would capture periodic screenshots of a user’s screen for later search. The concept sparked immediate privacy concerns, with critics warning that the feature could store sensitive data such as passwords and financial information. Microsoft delayed Recall for nearly a year to address security issues, but controversy persisted; a recent cybersecurity tool demonstrated that captured data could still be extracted, renewing questions about the fix.
Notion Mail
Notion Mail, an AI‑focused email product launched in April 2025, was intended to help users organize and automate their inboxes. The company discovered that customers were preferring separate AI agents for email management, leading to the decision to discontinue the service on September 22.
Humane AI Pin
The Humane AI Pin, a wearable device promising AI functionality without a smartphone, attracted $230 million in investment and significant media attention. The product suffered from performance problems, and a safety warning about a potential battery fire risk in its charging case further damaged its reputation. Humane ceased the AI Pin business in February 2025, and HP later acquired most of its assets for $116 million.
Rabbit R1
The Rabbit R1, introduced at CES in January 2024, was marketed as an AI companion capable of executing tasks on behalf of its owner. Early sales reached 100,000 units, but reviewers criticized the device as unfinished, with unreliable performance and limited integrations. Rabbit continues to update the hardware, now positioning it as a controller for agentic tasks, and has announced Project Cyberdeck, a new portable device aimed at vibe‑coding enthusiasts.
Huxe
Huxe, an AI audio application created by former Google NotebookLM developers, transformed written content into conversational podcast‑style audio. Competition from larger platforms, especially Spotify’s expanding AI audio tools, contributed to its closure in May 2026.
Yupp
Yupp offered a free platform where users could compare outputs from hundreds of AI models, vote on responses, and earn cryptocurrency. At its peak the service supported over 800 models from companies including OpenAI, Google, and Anthropic. The founders concluded that the product never achieved sufficient market fit, leading to its shutdown in March 2026.
Figgs AI
Figgs AI operated from 2023 to 2024, enabling users to create and interact with customizable AI characters for role‑play and storytelling. The platform attracted more than one million users, but sustaining a free service proved financially unsustainable, prompting its closure.
Character.ai
Character.ai, founded in 2021 by former Google researchers Noam Shazeer and Daniel De Freitas, gained popularity by allowing users to chat with customizable bots modeled after celebrities and fictional characters. The company raised $150 million at a $1 billion valuation in 2023, but Google later paid an estimated $2.7 billion to license its technology and bring the founders back, leaving the startup to continue independently. Subsequent legal challenges included lawsuits alleging that chatbots contributed to teen suicides and a Disney cease‑and‑desist over unauthorized character use. After restricting under‑18 users and settling several suits with Google in early 2026, Character.ai filed for bankruptcy in October, citing legal, regulatory, and operational pressures. Disney is now the lead bidder, offering $20 million for a license to some of its intellectual property, with the sale expected to conclude by late November or early December while the app remains operational.
