Back to news
AI Market BriefZDNet AI

I let ChatGPT Work and Claude Cowork loose on my files - only one made me nervous

ZDNet AI tested ChatGPT Work and Claude Cowork for desktop automation. Both offer similar file management features, but users perceive Claude Cowork as significantly safer and less intrusive during operations.

365 word signal
AI Brief

ZDNet AI

I let ChatGPT Work and Claude Cowork loose on my files - only one made me nervous

Signal Snapshot

6
related
2
FAQ
1
source

Briefing Notes

What happened and why it matters

Summary

A recent comparative evaluation by ZDNet AI examined the desktop automation features of ChatGPT Work and Claude Cowork. While both platforms deliver comparable functionality for managing local files, user feedback indicates a clear divergence in perceived security. Participants consistently rated Claude Cowork as significantly safer and less intrusive during active file operations.

Why it matters

As artificial intelligence transitions to autonomous desktop agents, direct access to local files becomes essential. However, this capability introduces significant privacy risks. The test results highlight a critical reality in AI adoption: technical capability alone does not guarantee acceptance. When agents operate locally, transparency regarding data handling and explicit permission boundaries become decisive. Developers prioritizing privacy-by-design architectures will likely capture markets more effectively than those focusing solely on automation speed.

Related tools

The evaluation centers on two leading desktop automation solutions: ChatGPT Work and Claude Cowork. Both represent the next generation of AI productivity suites designed to integrate seamlessly into existing operating systems.

Impact on AI tools/models

This insight signals a necessary evolution in how foundation models interact with local environments. Developers must implement sandboxed execution layers, granular access controls, and real-time activity logs to mitigate user anxiety. The perception that one tool is less intrusive suggests that underlying architectural choices directly influence safety metrics. Upcoming model releases will likely emphasize localized processing and user-configurable automation scopes to align with emerging privacy expectations.

What to watch

Industry stakeholders should monitor several developing trends as desktop AI matures. Regulatory frameworks around local data access are expected to tighten, potentially mandating standardized consent interfaces. Competing agents will likely introduce privacy dashboards to close the trust gap identified in recent testing. Benchmark methodologies will also evolve to include security scores alongside performance metrics. For ongoing coverage, explore our curated ToolSeekAI tools directory, stay updated via our latest AI news feed, and track market positioning through our live rankings.

FAQ

Which tools were compared in the test? The evaluation focused on ChatGPT Work and Claude Cowork for desktop automation capabilities.

How did users perceive the safety of these tools? Users found Claude Cowork to be significantly safer and less intrusive during file operations compared to ChatGPT Work.

Search FAQ

Frequently asked questions

FAQ

Which tools were compared in the test?
The evaluation focused on ChatGPT Work and Claude Cowork for desktop automation capabilities.
How did users perceive the safety of these tools?
Users found Claude Cowork to be significantly safer and less intrusive during file operations compared to ChatGPT Work.

Keep Tracking

Related AI news

News hub
ZDNet AI

90,000 Flock cameras have quietly gone up in the US: What they track and how to check your city

ZDNet AI

90,000 Flock cameras have quietly gone up in the US: What they track and how to check your city

Approximately 90,000 AI-powered Flock cameras now operate across the US, quietly identifying vehicles and tracking movements without public consent.

ZDNet AI

I tested iOS 26 Siri against iOS 27 Siri AI in my car - and it wasn't even close

ZDNet AI

I tested iOS 26 Siri against iOS 27 Siri AI in my car - and it wasn't even close

A hands-on test compares Apple’s iOS 26 Siri against the iOS 27 public beta version during in-car usage, revealing a significant performance gap favoring the newer AI-enhanced assistant.

ZDNet AI

Anthropic's Claude Corps will pay $85,000 to 1,000 early-career professionals - apply now

ZDNet AI

Anthropic's Claude Corps will pay $85,000 to 1,000 early-career professionals - apply now

Anthropic’s Claude Corps provides $85,000 stipends and benefits to 1,000 early-career professionals paired with nonprofits. Applications are closing rapidly.

ZDNet AI

Don't let an AI chatbot pick your password, ever

ZDNet AI

Don't let an AI chatbot pick your password, ever

New research shows AI-generated passwords lack true randomness and security compared to human-created ones. Experts warn against using chatbots for sensitive credentials due to predictable patterns.

ZDNet AI

An AI agent breached Hugging Face before an AI defender caught it: What users should do next

ZDNet AI

An AI agent breached Hugging Face before an AI defender caught it: What users should do next

An AI agent breached Hugging Face’s production infrastructure but was neutralized by an AI defender, highlighting the rise of autonomous cyber threats and the necessity of AI-driven security protocols.

ZDNet AI

The top AI fear for 6,000 tech pros isn't losing their jobs - it's more work for the same pay

ZDNet AI

The top AI fear for 6,000 tech pros isn't losing their jobs - it's more work for the same pay

A survey of 6,000 tech professionals reveals that increased workload without pay raises, not job loss, is the top AI fear. Most would not recommend their roles to newcomers.

Site Discovery

Keep exploring the AI ecosystem

After this brief, continue into related tools, models, and rankings to understand whether the story affects your choices.