Claude's new AI can now complete five-hour tasks

PLUS: OpenAI and Anthropic unite, and ChatGPT gets a shopping cart

Author

Gen AI Art
December 22, 2025 • Estimated Reading Time: 9 minutes

In partnership with

OpenAI has new research exploring whether an AI’s internal ‘thinking’ can be safely monitored. The findings suggest that tracking a model's step-by-step reasoning is an effective method for catching problematic behavior.

But this transparency isn't a given, as the study also found that models can be trained to deliberately hide their internal process. With that risk, is monitoring a reliable long-term safety mechanism, or just a temporary fix?

Today in AI:

  • Claude's new AI that completes five-hour tasks

  • OpenAI's test to read an AI's mind

  • ChatGPT gets a shopping cart

Turn AI Into Your Income Stream

The AI economy is booming, and smart entrepreneurs are already profiting. Subscribe to Mindstream and get instant access to 200+ proven strategies to monetize AI tools like ChatGPT, Midjourney, and more. From content creation to automation services, discover actionable ways to build your AI-powered income. No coding required, just practical strategies that work.

OpenAI Tests Reading an AI's Mind

What’s new? OpenAI published new research confirming that we can monitor an AI's step-by-step reasoning to catch issues, but this transparency isn't guaranteed as models advance.

What matters?

  • Monitoring an AI's "chain of thought" is dramatically more effective at spotting unwanted behavior than just looking at the final output, especially when the model produces longer reasoning.

  • There's a "monitorability tax"—using smaller, less complex models for a task can make them easier to monitor, but this often requires more compute to achieve the same performance as a larger model.

  • Researchers found models could be trained to deliberately hide their reasoning, making it crucial to actively preserve this transparency as AI scales; you can dive into the details in the full paper.

Why it matters?

If perfect AI alignment remains elusive, monitoring an AI's internal monologue becomes a critical safety mechanism for deploying advanced systems. This work signals that AI transparency must be an intentional engineering goal, not just a byproduct of model development.

GUIDE

ChatGPT gets a shopping cart

What’s new? ChatGPT now lets you order groceries directly through Instacart in a new partnership, turning meal ideas into a shopping cart without ever leaving the chat interface.

What matters?

  • The integration lets you handle the entire grocery shopping process—from finding items to final checkout—all within a single chat session.

  • This moves the AI beyond simply providing recipes; it now directly executes tasks by helping you purchase the necessary ingredients on the spot.

  • The feature builds on an existing relationship between the two companies, signaling a deeper push into integrating AI with real-world commerce.

Why it matters?

This integration shows how AI is evolving from an information tool into an action-oriented platform. It points toward a future where your AI assistant can manage complex, multi-step tasks for you through simple conversational commands.

PRESENTED BY GLADLY.AI

Your competitors are already automating. Here's the data.

Retail and ecommerce teams using AI for customer service are resolving 40-60% more tickets without more staff, cutting cost-per-ticket by 30%+, and handling seasonal spikes 3x faster.

But here's what separates winners from everyone else: they started with the data, not the hype.

Gladly handles the predictable volume, FAQs, routing, returns, order status, while your team focuses on customers who need a human touch. The result? Better experiences. Lower costs. Real competitive advantage. Ready to see what's possible for your business?

Claude comes to your code chat

What’s new? Anthropic just launched a new beta integration that brings Claude Code directly into Slack. This allows developer teams to turn chat threads into automated coding workflows without leaving their conversations.

What matters?

  • Tagging @Claude in a Slack thread creates a full coding session that uses the conversation's context, like bug reports or feature requests, as its starting point.

  • The integration handles the entire workflow by automatically selecting the right repository, posting progress updates, and delivering links to review changes and create pull requests.

  • This feature is a major expansion of Anthropic's existing Slack app, which previously offered only lightweight chat assistance.

Why it matters?

This integration meets developers where they already collaborate, turning the central communication hub into a powerful and interactive coding environment. It makes the AI feel less like a separate application and more like an embedded teammate, streamlining the path from discussion to deployment.

Everything else in AI

OpenAI hinted at its future roadmap, with mentions of incremental updates like GPT-5.1 and GPT-5.2 appearing in recent communications.

Artificial Analysis tracks the growing "monitorability tax," providing benchmarks on the increased inference compute costs of running more transparent but smaller AI models.

Researchers published a new paper on more interpretable AI, proposing a path toward systems whose decision-making processes are easier to understand and verify.

Essential AI Guides - Reading List:

Let us know!

Work with us

Reach 100k+ engaged Tech Professionals, Engineers, Managers and decision makers. Join brands like MorningBrew, HubSpot, Prezi, Nike, Ahref, Roku, 1440, Superhuman, and others in showcasing your product to our audience. Get in touch now →