Major AI Chatbots: OpenAI, Anthropic, Google Gemini, and the Road Ahead
OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, and Microsoft’s Copilot dominate the chatbot scene. But reports from The Atlantic, Business Insider, WIRED, Boing Boing, India Today, and IndiaVision show cracks in their operations, from IP fights to ballooning costs and new device pushes.
IP Rules That Favor the Big Players
OpenAI, Anthropic, Google, and xAI all block people from using their chatbot outputs to train rival models. That’s straight from their terms of service, as The Atlantic detailed in a March 20 piece. These same companies train their systems on books, videos, and other copyrighted works scraped from the web. OpenAI calls it “publicly available information.” Anthropic says it used books but not in commercial products. Google and others defend it as fair use.
Eric Schmidt, ex-Google CEO, told Stanford students in 2024 to grab data for a test AI and sort legal issues later if it succeeds. Anthropic’s CEO Dario Amodei wrote an internal 2021 memo pushing to pay creators a cut of profits to avoid backlash. Now the company fights authors in court over fair use. Meanwhile, these firms patent their own tech fiercely—Apple won over $1 billion from Samsung in a 2010s phone design fight, and Waymo settled with Uber for $245 million over self-driving secrets.
Token Costs Add Up Fast
Using ChatGPT, Claude, or xAI’s Grok isn’t cheap. Tokens—chunks of text that measure AI work—cost more for tougher tasks and top models. Box CEO Aaron Levie wrote on X that every worker will rack up bills, from engineers to legal and sales teams. Agents running overnight on huge data sets will burn the most.
Nvidia CEO Jensen Huang said he’d be alarmed if a $500,000 engineer spent just $5,000 on tokens. Venture capitalist Chamath Palihapitiya cut off Cursor AI coding tool use at his firm after costs tripled since late 2025, hitting millions across AWS, Cursor, and Anthropic. Levie expects business units, maybe even CFOs, to own these budgets.
AI Agents Push into Phones
Google’s Gemini now handles tasks like ordering Uber rides or DoorDash on Samsung and Pixel phones. As WIRED reported on March 20, rumors point to an Amazon “Transformer” phone with Alexa+ AI for shopping, possibly skipping app stores via generated interfaces. OpenAI teams with ex-Apple designer Jony Ive on AI devices that act as collaborators.
Other firms eye similar shifts. T-Mobile’s parent showed a concept phone at Mobile World Congress 2024 that builds interfaces from speech. Nothing’s CEO predicts one app as the whole OS.
The Road Ahead
Current chatbots still hit walls on memory and retrieval. Boing Boing described how a startup shells out $800 a day to expose flaws in major models.
Privacy draws fire too. India Today reports Mark Zuckerberg linking with Signal’s founder for end-to-end encrypted chats in Meta AI.
Harms spark lawsuits. Families fight to make AI firms answer for children’s deaths tied to chatbot talks, IndiaVision notes.
- OpenAI/ChatGPT: Trains on public data, bans output use for rivals, token costs rising.
- Anthropic/Claude: Similar IP bans and fair use claims, heavy token spender.
- Google/Gemini: IP protector, agent tasks on phones.
- Microsoft Copilot: Builds on OpenAI tech amid these trends.
