← Back to archive
ChatGPT's mobile boom is already over

Image via Ars Technica

Tuesday, October 28, 2025

ChatGPT's mobile boom is already over

The ChatGPT mobile app is hitting a growth plateau (yikes) as users move past the novelty phase, while Anthropic boldly pushes Claude into financial services with Excel integration and live data capabilities. Meanwhile, researchers discovered AI search engines are actually favoring obscure sources over popular websites, and OpenAI just boosted ChatGPT's mental health safety by 65-80% with expert training. Here's the real question: If ChatGPT's growth is slowing on mobile, are we hitting peak consumer AI saturation?

Top Stories

1
ChatGPT's Mobile App Is Seeing Slowing Download Growth and Daily Use, Analysis Shows

TechCrunch

ChatGPT's mobile app growth has peaked with declining downloads, daily active users, and engagement metrics, signaling the end of its novelty phase and requiring OpenAI to invest in new features and marketing to sustain momentum.

chatgptmobileopenaicompetition
2
Advancing Claude for Financial Services

Anthropic

Anthropic launches Claude for Financial Services with Excel integration, live market data connectors, and specialized AI agent skills, enabling major financial institutions to automate complex analytical tasks and significantly boost productivity.

claudeanthropicenterprise-aifinancial-services
3
AI-Powered Search Engines Rely on "Less Popular" Sources, Researchers Find

Ars Technica

AI-powered search engines systematically cite less popular and niche websites compared to traditional search results, with over half of sources appearing nowhere near Google's organic top results, raising questions about source reliability and information discovery patterns.

searchai-overviewsgooglegemini
4
Strengthening ChatGPT's Responses in Sensitive Conversations

OpenAI

OpenAI released safety improvements to ChatGPT's default model that reduce harmful responses in mental health conversations by 65-80%, developed with 170+ clinicians and validated through expert reviews and rigorous testing protocols.

openaisafetymental-healthresponsible-ai
5
Speedrunning an RL Environment

Building RL environments for LLM training requires understanding environment primitives, state management, and reward structures; this guide demonstrates adapting the AgentDojo security benchmark using the verifiers framework while identifying key performance bottlenecks and architectural considerations.

rl-environmentsllm-trainingagentsprompt-injection

Keep Reading

Industry Voices

Enjoyed this issue?

Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.