Character.ai Goes Bankrupt: TechCrunch Builds a Graveyard of AI Projects From OpenAI and Apple

⚡ The gist

• Character.ai filed for bankruptcy in early October 2026 — five years after launch.
• Per S&P Global Market Intelligence, roughly 42% of corporate AI initiatives end up abandoned.
• OpenAI rolled back its "superapp" with Chat, Codex, and Work modes; the Atlas browser and the Sora video platform are gone.

🔍 What happened

On November 14, TechCrunch published a running list of AI projects that shut down, wound down, or failed to deliver on their promises. The latest entries: Character.ai's bankruptcy in early October and the closure of the privacy assistant Ollie a few weeks earlier. For scale: S&P Global Market Intelligence estimates that about 42% of corporate AI initiatives ultimately get abandoned — due to lack of funding, technical hurdles, competition, scaling problems, or weak demand.

Character.ai has its own timeline. The 2021 startup was founded by Google alumni Noam Shazeer and Daniel De Freitas; in 2023 it raised $150 million at a $1 billion valuation. A year later, Google reportedly paid $2.7 billion to license the technology and bring the founders back. Then came lawsuits from families claiming the bots had driven teenagers to suicide and self-harm, a demand from Disney to stop using its characters, a ban on open chats for users under 18, and settlements — including with Google — in early 2026. The October bankruptcy was attributed to legal, regulatory, and operational problems.

OpenAI took a different path — through rollbacks. On July 9, the app was reworked into a "superapp" with separate Chat, Codex, and Work modes, while the familiar version was rebranded ChatGPT Classic. The interface was panned as cluttered, and the update was rolled back. The ChatGPT Atlas browser lasted less than a year and shut down on August 9, with its features absorbed into ChatGPT. The Operator agent's capabilities moved there too, and DALL-E shrank as a standalone entry point. The Sora video platform died in March 2026: too expensive to run, weak retention.

Apple dragged its feet on Siri. The new version was promised back in 2024 alongside Apple Intelligence. The release kept slipping due to engineering problems and bugs — and it cost the company $250 million in a settlement over lawsuits about how it advertised the iPhone 16's AI features. Siri AI reached the iOS 27 beta in July, and this month it started rolling out to users — English speakers first.

🧠 Our take

The list didn't come out because of one event — it's an accumulated tally. Our read: 2026 was the year the money for experiments ran out. S&P's 42% figure isn't about the technology; it's about the economics — expensive to run, weak retention, and investors doing the math. Sora, with its high operating costs, is exactly that scenario, just played out in public.

What's changing: the industry is moving away from measuring success by the number of launches. Look at OpenAI — standalone apps are collapsing back into ChatGPT, and the source itself says that strategy worked better than the "superapp" redesign. For your wallet, that means fewer subscriptions and fewer reasons to install a zoo of clients. For users, one front door instead of a dozen.

Who wins: platforms with somewhere to embed AI — Google, Apple, OpenAI itself. Who loses: niche startups without distribution. Character.ai is the cautionary tale: a $1 billion valuation, $2.7 billion from Google for the license — and bankruptcy at the end of the day. We think the "build a standalone AI product" model, without a sales channel, will die faster than people expect.

⚠️ What we don't know

The list is a journalistic selection: TechCrunch doesn't set inclusion criteria, so the "graveyard" is neither complete nor rigorous. S&P's estimate is a single number, with no breakdown by industry or company size. For Character.ai's bankruptcy, no filing procedure has been named and there are no financial details. We haven't verified every item on the list against primary documents — worth doing yourself if a decision hinges on these figures.

Forecast: within the next quarter, TechCrunch's list will gain at least one more entry about a major player shutting down a standalone AI app — the market is pushing products into core services, not standalone brands.

Disclaimer: single source — a TechCrunch roundup; individual items on the list are not backed by primary documents.

Source: TechCrunch